]> git.ipfire.org Git - thirdparty/kernel/stable.git/blame - net/ipv6/route.c
projects / thirdparty / kernel / stable.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
Merge branch 'tcp_ack_loops'
[thirdparty/kernel/stable.git] / net / ipv6 / route.c
CommitLineData
1da177e4
LT		 1/*
2 * Linux INET6 implementation
3 * FIB front-end.
4 *
5 * Authors:
1ab1457c 6 * Pedro Roque <roque@di.fc.ul.pt>
1da177e4 7 *
1da177e4
LT		 8 * This program is free software; you can redistribute it and/or
9 * modify it under the terms of the GNU General Public License
10 * as published by the Free Software Foundation; either version
11 * 2 of the License, or (at your option) any later version.
12 */
13
14/* Changes:
15 *
16 * YOSHIFUJI Hideaki @USAGI
17 * reworked default router selection.
18 * - respect outgoing interface
19 * - select from (probably) reachable routers (i.e.
20 * routers in REACHABLE, STALE, DELAY or PROBE states).
21 * - always select the same router if it is (probably)
22 * reachable. otherwise, round-robin the list.
c0bece9f
YH		 23 * Ville Nuorvala
24 * Fixed routing subtrees.
1da177e4
LT		 25 */
26
f3213831
JP		 27#define pr_fmt(fmt) "IPv6: " fmt
28
4fc268d2 29#include <linux/capability.h>
1da177e4 30#include <linux/errno.h>
bc3b2d7f 31#include <linux/export.h>
1da177e4
LT		 32#include <linux/types.h>
33#include <linux/times.h>
34#include <linux/socket.h>
35#include <linux/sockios.h>
36#include <linux/net.h>
37#include <linux/route.h>
38#include <linux/netdevice.h>
39#include <linux/in6.h>
7bc570c8 40#include <linux/mroute6.h>
1da177e4 41#include <linux/init.h>
1da177e4 42#include <linux/if_arp.h>
1da177e4
LT		 43#include <linux/proc_fs.h>
44#include <linux/seq_file.h>
5b7c931d 45#include <linux/nsproxy.h>
5a0e3ad6 46#include <linux/slab.h>
457c4cbc 47#include <net/net_namespace.h>
1da177e4
LT		 48#include <net/snmp.h>
49#include <net/ipv6.h>
50#include <net/ip6_fib.h>
51#include <net/ip6_route.h>
52#include <net/ndisc.h>
53#include <net/addrconf.h>
54#include <net/tcp.h>
55#include <linux/rtnetlink.h>
56#include <net/dst.h>
57#include <net/xfrm.h>
8d71740c 58#include <net/netevent.h>
21713ebc 59#include <net/netlink.h>
51ebd318 60#include <net/nexthop.h>
1da177e4
LT		 61
62#include <asm/uaccess.h>
63
64#ifdef CONFIG_SYSCTL
65#include <linux/sysctl.h>
66#endif
67
afc154e9 68enum rt6_nud_state {
7e980569
JB		 69 RT6_NUD_FAIL_HARD = -3,
70 RT6_NUD_FAIL_PROBE = -2,
71 RT6_NUD_FAIL_DO_RR = -1,
afc154e9
HFS		 72 RT6_NUD_SUCCEED = 1
73};
74
1716a961 75static struct rt6_info *ip6_rt_copy(struct rt6_info *ort,
21efcfa0 76 const struct in6_addr *dest);
1da177e4 77static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie);
0dbaee3b 78static unsigned int ip6_default_advmss(const struct dst_entry *dst);
ebb762f2 79static unsigned int ip6_mtu(const struct dst_entry *dst);
1da177e4
LT		 80static struct dst_entry *ip6_negative_advice(struct dst_entry *);
81static void ip6_dst_destroy(struct dst_entry *);
82static void ip6_dst_ifdown(struct dst_entry *,
83 struct net_device *dev, int how);
569d3645 84static int ip6_dst_gc(struct dst_ops *ops);
1da177e4
LT		 85
86static int ip6_pkt_discard(struct sk_buff *skb);
aad88724 87static int ip6_pkt_discard_out(struct sock *sk, struct sk_buff *skb);
7150aede 88static int ip6_pkt_prohibit(struct sk_buff *skb);
aad88724 89static int ip6_pkt_prohibit_out(struct sock *sk, struct sk_buff *skb);
1da177e4 90static void ip6_link_failure(struct sk_buff *skb);
6700c270
DM		 91static void ip6_rt_update_pmtu(struct dst_entry *dst, struct sock *sk,
92 struct sk_buff *skb, u32 mtu);
93static void rt6_do_redirect(struct dst_entry *dst, struct sock *sk,
94 struct sk_buff *skb);
52bd4c0c 95static int rt6_score_route(struct rt6_info *rt, int oif, int strict);
1da177e4 96
70ceb4f5 97#ifdef CONFIG_IPV6_ROUTE_INFO
efa2cea0 98static struct rt6_info *rt6_add_route_info(struct net *net,
b71d1d42
ED		 99 const struct in6_addr *prefix, int prefixlen,
100 const struct in6_addr *gwaddr, int ifindex,
95c96174 101 unsigned int pref);
efa2cea0 102static struct rt6_info *rt6_get_route_info(struct net *net,
b71d1d42
ED		 103 const struct in6_addr *prefix, int prefixlen,
104 const struct in6_addr *gwaddr, int ifindex);
70ceb4f5
YH		 105#endif
106
e8243534 107static void rt6_bind_peer(struct rt6_info *rt, int create)
108{
109 struct inet_peer_base *base;
110 struct inet_peer *peer;
111
112 base = inetpeer_base_ptr(rt->_rt6i_peer);
113 if (!base)
114 return;
115
116 peer = inet_getpeer_v6(base, &rt->rt6i_dst.addr, create);
117 if (peer) {
118 if (!rt6_set_peer(rt, peer))
119 inet_putpeer(peer);
120 }
121}
122
123static struct inet_peer *__rt6_get_peer(struct rt6_info *rt, int create)
124{
125 if (rt6_has_peer(rt))
126 return rt6_peer_ptr(rt);
127
128 rt6_bind_peer(rt, create);
129 return (rt6_has_peer(rt) ? rt6_peer_ptr(rt) : NULL);
130}
131
132static struct inet_peer *rt6_get_peer_create(struct rt6_info *rt)
133{
134 return __rt6_get_peer(rt, 1);
135}
136
06582540
DM		 137static u32 *ipv6_cow_metrics(struct dst_entry *dst, unsigned long old)
138{
139 struct rt6_info *rt = (struct rt6_info *) dst;
140 struct inet_peer *peer;
141 u32 *p = NULL;
142
8e2ec639
YZ		 143 if (!(rt->dst.flags & DST_HOST))
144 return NULL;
145
fbfe95a4 146 peer = rt6_get_peer_create(rt);
06582540
DM		 147 if (peer) {
148 u32 *old_p = __DST_METRICS_PTR(old);
149 unsigned long prev, new;
150
151 p = peer->metrics;
e5fd387a
MK		 152 if (inet_metrics_new(peer) ||
153 (old & DST_METRICS_FORCE_OVERWRITE))
06582540
DM		 154 memcpy(p, old_p, sizeof(u32) * RTAX_MAX);
155
156 new = (unsigned long) p;
157 prev = cmpxchg(&dst->_metrics, old, new);
158
159 if (prev != old) {
160 p = __DST_METRICS_PTR(prev);
161 if (prev & DST_METRICS_READ_ONLY)
162 p = NULL;
163 }
164 }
165 return p;
166}
167
f894cbf8
DM		 168static inline const void *choose_neigh_daddr(struct rt6_info *rt,
169 struct sk_buff *skb,
170 const void *daddr)
39232973
DM		 171{
172 struct in6_addr *p = &rt->rt6i_gateway;
173
a7563f34 174 if (!ipv6_addr_any(p))
39232973 175 return (const void *) p;
f894cbf8
DM		 176 else if (skb)
177 return &ipv6_hdr(skb)->daddr;
39232973
DM		 178 return daddr;
179}
180
f894cbf8
DM		 181static struct neighbour *ip6_neigh_lookup(const struct dst_entry *dst,
182 struct sk_buff *skb,
183 const void *daddr)
d3aaeb38 184{
39232973
DM		 185 struct rt6_info *rt = (struct rt6_info *) dst;
186 struct neighbour *n;
187
f894cbf8 188 daddr = choose_neigh_daddr(rt, skb, daddr);
8e022ee6 189 n = __ipv6_neigh_lookup(dst->dev, daddr);
f83c7790
DM		 190 if (n)
191 return n;
192 return neigh_create(&nd_tbl, daddr, dst->dev);
193}
194
9a7ec3a9 195static struct dst_ops ip6_dst_ops_template = {
1da177e4 196 .family = AF_INET6,
09640e63 197 .protocol = cpu_to_be16(ETH_P_IPV6),
1da177e4
LT		 198 .gc = ip6_dst_gc,
199 .gc_thresh = 1024,
200 .check = ip6_dst_check,
0dbaee3b 201 .default_advmss = ip6_default_advmss,
ebb762f2 202 .mtu = ip6_mtu,
06582540 203 .cow_metrics = ipv6_cow_metrics,
1da177e4
LT		 204 .destroy = ip6_dst_destroy,
205 .ifdown = ip6_dst_ifdown,
206 .negative_advice = ip6_negative_advice,
207 .link_failure = ip6_link_failure,
208 .update_pmtu = ip6_rt_update_pmtu,
6e157b6a 209 .redirect = rt6_do_redirect,
1ac06e03 210 .local_out = __ip6_local_out,
d3aaeb38 211 .neigh_lookup = ip6_neigh_lookup,
1da177e4
LT		 212};
213
ebb762f2 214static unsigned int ip6_blackhole_mtu(const struct dst_entry *dst)
ec831ea7 215{
618f9bc7
SK		 216 unsigned int mtu = dst_metric_raw(dst, RTAX_MTU);
217
218 return mtu ? : dst->dev->mtu;
ec831ea7
RD		 219}
220
6700c270
DM		 221static void ip6_rt_blackhole_update_pmtu(struct dst_entry *dst, struct sock *sk,
222 struct sk_buff *skb, u32 mtu)
14e50e57
DM		 223{
224}
225
6700c270
DM		 226static void ip6_rt_blackhole_redirect(struct dst_entry *dst, struct sock *sk,
227 struct sk_buff *skb)
b587ee3b
DM		 228{
229}
230
0972ddb2
HB		 231static u32 *ip6_rt_blackhole_cow_metrics(struct dst_entry *dst,
232 unsigned long old)
233{
234 return NULL;
235}
236
14e50e57
DM		 237static struct dst_ops ip6_dst_blackhole_ops = {
238 .family = AF_INET6,
09640e63 239 .protocol = cpu_to_be16(ETH_P_IPV6),
14e50e57
DM		 240 .destroy = ip6_dst_destroy,
241 .check = ip6_dst_check,
ebb762f2 242 .mtu = ip6_blackhole_mtu,
214f45c9 243 .default_advmss = ip6_default_advmss,
14e50e57 244 .update_pmtu = ip6_rt_blackhole_update_pmtu,
b587ee3b 245 .redirect = ip6_rt_blackhole_redirect,
0972ddb2 246 .cow_metrics = ip6_rt_blackhole_cow_metrics,
d3aaeb38 247 .neigh_lookup = ip6_neigh_lookup,
14e50e57
DM		 248};
249
62fa8a84 250static const u32 ip6_template_metrics[RTAX_MAX] = {
14edd87d 251 [RTAX_HOPLIMIT - 1] = 0,
62fa8a84
DM		 252};
253
fb0af4c7 254static const struct rt6_info ip6_null_entry_template = {
d8d1f30b
CG		 255 .dst = {
256 .__refcnt = ATOMIC_INIT(1),
257 .__use = 1,
2c20cbd7 258 .obsolete = DST_OBSOLETE_FORCE_CHK,
d8d1f30b 259 .error = -ENETUNREACH,
d8d1f30b
CG		 260 .input = ip6_pkt_discard,
261 .output = ip6_pkt_discard_out,
1da177e4
LT		 262 },
263 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
4f724279 264 .rt6i_protocol = RTPROT_KERNEL,
1da177e4
LT		 265 .rt6i_metric = ~(u32) 0,
266 .rt6i_ref = ATOMIC_INIT(1),
267};
268
101367c2
TG		 269#ifdef CONFIG_IPV6_MULTIPLE_TABLES
270
fb0af4c7 271static const struct rt6_info ip6_prohibit_entry_template = {
d8d1f30b
CG		 272 .dst = {
273 .__refcnt = ATOMIC_INIT(1),
274 .__use = 1,
2c20cbd7 275 .obsolete = DST_OBSOLETE_FORCE_CHK,
d8d1f30b 276 .error = -EACCES,
d8d1f30b
CG		 277 .input = ip6_pkt_prohibit,
278 .output = ip6_pkt_prohibit_out,
101367c2
TG		 279 },
280 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
4f724279 281 .rt6i_protocol = RTPROT_KERNEL,
101367c2
TG		 282 .rt6i_metric = ~(u32) 0,
283 .rt6i_ref = ATOMIC_INIT(1),
284};
285
fb0af4c7 286static const struct rt6_info ip6_blk_hole_entry_template = {
d8d1f30b
CG		 287 .dst = {
288 .__refcnt = ATOMIC_INIT(1),
289 .__use = 1,
2c20cbd7 290 .obsolete = DST_OBSOLETE_FORCE_CHK,
d8d1f30b 291 .error = -EINVAL,
d8d1f30b 292 .input = dst_discard,
aad88724 293 .output = dst_discard_sk,
101367c2
TG		 294 },
295 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
4f724279 296 .rt6i_protocol = RTPROT_KERNEL,
101367c2
TG		 297 .rt6i_metric = ~(u32) 0,
298 .rt6i_ref = ATOMIC_INIT(1),
299};
300
301#endif
302
1da177e4 303/* allocate dst with ip6_dst_ops */
97bab73f 304static inline struct rt6_info *ip6_dst_alloc(struct net *net,
957c665f 305 struct net_device *dev,
8b96d22d
DM		 306 int flags,
307 struct fib6_table *table)
1da177e4 308{
97bab73f 309 struct rt6_info *rt = dst_alloc(&net->ipv6.ip6_dst_ops, dev,
6f3118b5 310 0, DST_OBSOLETE_FORCE_CHK, flags);
cf911662 311
97bab73f 312 if (rt) {
8104891b
SK		 313 struct dst_entry *dst = &rt->dst;
314
315 memset(dst + 1, 0, sizeof(*rt) - sizeof(*dst));
8b96d22d 316 rt6_init_peer(rt, table ? &table->tb6_peers : net->ipv6.peers);
51ebd318 317 INIT_LIST_HEAD(&rt->rt6i_siblings);
97bab73f 318 }
cf911662 319 return rt;
1da177e4
LT		 320}
321
322static void ip6_dst_destroy(struct dst_entry *dst)
323{
324 struct rt6_info *rt = (struct rt6_info *)dst;
325 struct inet6_dev *idev = rt->rt6i_idev;
ecd98837 326 struct dst_entry *from = dst->from;
1da177e4 327
8e2ec639
YZ		 328 if (!(rt->dst.flags & DST_HOST))
329 dst_destroy_metrics_generic(dst);
330
38308473 331 if (idev) {
1da177e4
LT		 332 rt->rt6i_idev = NULL;
333 in6_dev_put(idev);
1ab1457c 334 }
1716a961 335
ecd98837
YH		 336 dst->from = NULL;
337 dst_release(from);
1716a961 338
97bab73f
DM		 339 if (rt6_has_peer(rt)) {
340 struct inet_peer *peer = rt6_peer_ptr(rt);
b3419363
DM		 341 inet_putpeer(peer);
342 }
343}
344
1da177e4
LT		 345static void ip6_dst_ifdown(struct dst_entry *dst, struct net_device *dev,
346 int how)
347{
348 struct rt6_info *rt = (struct rt6_info *)dst;
349 struct inet6_dev *idev = rt->rt6i_idev;
5a3e55d6 350 struct net_device *loopback_dev =
c346dca1 351 dev_net(dev)->loopback_dev;
1da177e4 352
97cac082
DM		 353 if (dev != loopback_dev) {
354 if (idev && idev->dev == dev) {
355 struct inet6_dev *loopback_idev =
356 in6_dev_get(loopback_dev);
357 if (loopback_idev) {
358 rt->rt6i_idev = loopback_idev;
359 in6_dev_put(idev);
360 }
361 }
1da177e4
LT		 362 }
363}
364
a50feda5 365static bool rt6_check_expired(const struct rt6_info *rt)
1da177e4 366{
1716a961
G		 367 if (rt->rt6i_flags & RTF_EXPIRES) {
368 if (time_after(jiffies, rt->dst.expires))
a50feda5 369 return true;
1716a961 370 } else if (rt->dst.from) {
3fd91fb3 371 return rt6_check_expired((struct rt6_info *) rt->dst.from);
1716a961 372 }
a50feda5 373 return false;
1da177e4
LT		 374}
375
51ebd318
ND		 376/* Multipath route selection:
377 * Hash based function using packet header and flowlabel.
378 * Adapted from fib_info_hashfn()
379 */
380static int rt6_info_hash_nhsfn(unsigned int candidate_count,
381 const struct flowi6 *fl6)
382{
383 unsigned int val = fl6->flowi6_proto;
384
c08977bb
YH		 385 val ^= ipv6_addr_hash(&fl6->daddr);
386 val ^= ipv6_addr_hash(&fl6->saddr);
51ebd318
ND		 387
388 /* Work only if this not encapsulated */
389 switch (fl6->flowi6_proto) {
390 case IPPROTO_UDP:
391 case IPPROTO_TCP:
392 case IPPROTO_SCTP:
b3ce5ae1
ND		 393 val ^= (__force u16)fl6->fl6_sport;
394 val ^= (__force u16)fl6->fl6_dport;
51ebd318
ND		 395 break;
396
397 case IPPROTO_ICMPV6:
b3ce5ae1
ND		 398 val ^= (__force u16)fl6->fl6_icmp_type;
399 val ^= (__force u16)fl6->fl6_icmp_code;
51ebd318
ND		 400 break;
401 }
402 /* RFC6438 recommands to use flowlabel */
b3ce5ae1 403 val ^= (__force u32)fl6->flowlabel;
51ebd318
ND		 404
405 /* Perhaps, we need to tune, this function? */
406 val = val ^ (val >> 7) ^ (val >> 12);
407 return val % candidate_count;
408}
409
410static struct rt6_info *rt6_multipath_select(struct rt6_info *match,
52bd4c0c
ND		 411 struct flowi6 *fl6, int oif,
412 int strict)
51ebd318
ND		 413{
414 struct rt6_info *sibling, *next_sibling;
415 int route_choosen;
416
417 route_choosen = rt6_info_hash_nhsfn(match->rt6i_nsiblings + 1, fl6);
418 /* Don't change the route, if route_choosen == 0
419 * (siblings does not include ourself)
420 */
421 if (route_choosen)
422 list_for_each_entry_safe(sibling, next_sibling,
423 &match->rt6i_siblings, rt6i_siblings) {
424 route_choosen--;
425 if (route_choosen == 0) {
52bd4c0c
ND		 426 if (rt6_score_route(sibling, oif, strict) < 0)
427 break;
51ebd318
ND		 428 match = sibling;
429 break;
430 }
431 }
432 return match;
433}
434
1da177e4 435/*
c71099ac 436 * Route lookup. Any table->tb6_lock is implied.
1da177e4
LT		 437 */
438
8ed67789
DL		 439static inline struct rt6_info *rt6_device_match(struct net *net,
440 struct rt6_info *rt,
b71d1d42 441 const struct in6_addr *saddr,
1da177e4 442 int oif,
d420895e 443 int flags)
1da177e4
LT		 444{
445 struct rt6_info *local = NULL;
446 struct rt6_info *sprt;
447
dd3abc4e
YH		 448 if (!oif && ipv6_addr_any(saddr))
449 goto out;
450
d8d1f30b 451 for (sprt = rt; sprt; sprt = sprt->dst.rt6_next) {
d1918542 452 struct net_device *dev = sprt->dst.dev;
dd3abc4e
YH		 453
454 if (oif) {
1da177e4
LT		 455 if (dev->ifindex == oif)
456 return sprt;
457 if (dev->flags & IFF_LOOPBACK) {
38308473 458 if (!sprt->rt6i_idev ||
1da177e4 459 sprt->rt6i_idev->dev->ifindex != oif) {
d420895e 460 if (flags & RT6_LOOKUP_F_IFACE && oif)
1da177e4 461 continue;
1ab1457c 462 if (local && (!oif ||
1da177e4
LT		 463 local->rt6i_idev->dev->ifindex == oif))
464 continue;
465 }
466 local = sprt;
467 }
dd3abc4e
YH		 468 } else {
469 if (ipv6_chk_addr(net, saddr, dev,
470 flags & RT6_LOOKUP_F_IFACE))
471 return sprt;
1da177e4 472 }
dd3abc4e 473 }
1da177e4 474
dd3abc4e 475 if (oif) {
1da177e4
LT		 476 if (local)
477 return local;
478
d420895e 479 if (flags & RT6_LOOKUP_F_IFACE)
8ed67789 480 return net->ipv6.ip6_null_entry;
1da177e4 481 }
dd3abc4e 482out:
1da177e4
LT		 483 return rt;
484}
485
27097255 486#ifdef CONFIG_IPV6_ROUTER_PREF
c2f17e82
HFS		 487struct __rt6_probe_work {
488 struct work_struct work;
489 struct in6_addr target;
490 struct net_device *dev;
491};
492
493static void rt6_probe_deferred(struct work_struct *w)
494{
495 struct in6_addr mcaddr;
496 struct __rt6_probe_work *work =
497 container_of(w, struct __rt6_probe_work, work);
498
499 addrconf_addr_solict_mult(&work->target, &mcaddr);
500 ndisc_send_ns(work->dev, NULL, &work->target, &mcaddr, NULL);
501 dev_put(work->dev);
502 kfree(w);
503}
504
27097255
YH		 505static void rt6_probe(struct rt6_info *rt)
506{
f2c31e32 507 struct neighbour *neigh;
27097255
YH		 508 /*
509 * Okay, this does not seem to be appropriate
510 * for now, however, we need to check if it
511 * is really so; aka Router Reachability Probing.
512 *
513 * Router Reachability Probe MUST be rate-limited
514 * to no more than one per minute.
515 */
2152caea 516 if (!rt || !(rt->rt6i_flags & RTF_GATEWAY))
7ff74a59 517 return;
2152caea
YH		 518 rcu_read_lock_bh();
519 neigh = __ipv6_neigh_lookup_noref(rt->dst.dev, &rt->rt6i_gateway);
520 if (neigh) {
521 write_lock(&neigh->lock);
522 if (neigh->nud_state & NUD_VALID)
523 goto out;
7ff74a59 524 }
2152caea
YH		 525
526 if (!neigh ||
52e16356 527 time_after(jiffies, neigh->updated + rt->rt6i_idev->cnf.rtr_probe_interval)) {
c2f17e82 528 struct __rt6_probe_work *work;
27097255 529
c2f17e82
HFS		 530 work = kmalloc(sizeof(*work), GFP_ATOMIC);
531
532 if (neigh && work)
7e980569 533 __neigh_set_probe_once(neigh);
c2f17e82
HFS		 534
535 if (neigh)
2152caea
YH		 536 write_unlock(&neigh->lock);
537
c2f17e82
HFS		 538 if (work) {
539 INIT_WORK(&work->work, rt6_probe_deferred);
540 work->target = rt->rt6i_gateway;
541 dev_hold(rt->dst.dev);
542 work->dev = rt->dst.dev;
543 schedule_work(&work->work);
544 }
f2c31e32 545 } else {
2152caea
YH		 546out:
547 write_unlock(&neigh->lock);
f2c31e32 548 }
2152caea 549 rcu_read_unlock_bh();
27097255
YH		 550}
551#else
552static inline void rt6_probe(struct rt6_info *rt)
553{
27097255
YH		 554}
555#endif
556
1da177e4 557/*
554cfb7e 558 * Default Router Selection (RFC 2461 6.3.6)
1da177e4 559 */
b6f99a21 560static inline int rt6_check_dev(struct rt6_info *rt, int oif)
554cfb7e 561{
d1918542 562 struct net_device *dev = rt->dst.dev;
161980f4 563 if (!oif || dev->ifindex == oif)
554cfb7e 564 return 2;
161980f4
DM		 565 if ((dev->flags & IFF_LOOPBACK) &&
566 rt->rt6i_idev && rt->rt6i_idev->dev->ifindex == oif)
567 return 1;
568 return 0;
554cfb7e 569}
1da177e4 570
afc154e9 571static inline enum rt6_nud_state rt6_check_neigh(struct rt6_info *rt)
1da177e4 572{
f2c31e32 573 struct neighbour *neigh;
afc154e9 574 enum rt6_nud_state ret = RT6_NUD_FAIL_HARD;
f2c31e32 575
4d0c5911
YH		 576 if (rt->rt6i_flags & RTF_NONEXTHOP ||
577 !(rt->rt6i_flags & RTF_GATEWAY))
afc154e9 578 return RT6_NUD_SUCCEED;
145a3621
YH		 579
580 rcu_read_lock_bh();
581 neigh = __ipv6_neigh_lookup_noref(rt->dst.dev, &rt->rt6i_gateway);
582 if (neigh) {
583 read_lock(&neigh->lock);
554cfb7e 584 if (neigh->nud_state & NUD_VALID)
afc154e9 585 ret = RT6_NUD_SUCCEED;
398bcbeb 586#ifdef CONFIG_IPV6_ROUTER_PREF
a5a81f0b 587 else if (!(neigh->nud_state & NUD_FAILED))
afc154e9 588 ret = RT6_NUD_SUCCEED;
7e980569
JB		 589 else
590 ret = RT6_NUD_FAIL_PROBE;
398bcbeb 591#endif
145a3621 592 read_unlock(&neigh->lock);
afc154e9
HFS		 593 } else {
594 ret = IS_ENABLED(CONFIG_IPV6_ROUTER_PREF) ?
7e980569 595 RT6_NUD_SUCCEED : RT6_NUD_FAIL_DO_RR;
a5a81f0b 596 }
145a3621
YH		 597 rcu_read_unlock_bh();
598
a5a81f0b 599 return ret;
1da177e4
LT		 600}
601
554cfb7e
YH		 602static int rt6_score_route(struct rt6_info *rt, int oif,
603 int strict)
1da177e4 604{
a5a81f0b 605 int m;
1ab1457c 606
4d0c5911 607 m = rt6_check_dev(rt, oif);
77d16f45 608 if (!m && (strict & RT6_LOOKUP_F_IFACE))
afc154e9 609 return RT6_NUD_FAIL_HARD;
ebacaaa0
YH		 610#ifdef CONFIG_IPV6_ROUTER_PREF
611 m |= IPV6_DECODE_PREF(IPV6_EXTRACT_PREF(rt->rt6i_flags)) << 2;
612#endif
afc154e9
HFS		 613 if (strict & RT6_LOOKUP_F_REACHABLE) {
614 int n = rt6_check_neigh(rt);
615 if (n < 0)
616 return n;
617 }
554cfb7e
YH		 618 return m;
619}
620
f11e6659 621static struct rt6_info *find_match(struct rt6_info *rt, int oif, int strict,
afc154e9
HFS		 622 int *mpri, struct rt6_info *match,
623 bool *do_rr)
554cfb7e 624{
f11e6659 625 int m;
afc154e9 626 bool match_do_rr = false;
f11e6659
DM		 627
628 if (rt6_check_expired(rt))
629 goto out;
630
631 m = rt6_score_route(rt, oif, strict);
7e980569 632 if (m == RT6_NUD_FAIL_DO_RR) {
afc154e9
HFS		 633 match_do_rr = true;
634 m = 0; /* lowest valid score */
7e980569 635 } else if (m == RT6_NUD_FAIL_HARD) {
f11e6659 636 goto out;
afc154e9
HFS		 637 }
638
639 if (strict & RT6_LOOKUP_F_REACHABLE)
640 rt6_probe(rt);
f11e6659 641
7e980569 642 /* note that m can be RT6_NUD_FAIL_PROBE at this point */
f11e6659 643 if (m > *mpri) {
afc154e9 644 *do_rr = match_do_rr;
f11e6659
DM		 645 *mpri = m;
646 match = rt;
f11e6659 647 }
f11e6659
DM		 648out:
649 return match;
650}
651
652static struct rt6_info *find_rr_leaf(struct fib6_node *fn,
653 struct rt6_info *rr_head,
afc154e9
HFS		 654 u32 metric, int oif, int strict,
655 bool *do_rr)
f11e6659
DM		 656{
657 struct rt6_info *rt, *match;
554cfb7e 658 int mpri = -1;
1da177e4 659
f11e6659
DM		 660 match = NULL;
661 for (rt = rr_head; rt && rt->rt6i_metric == metric;
d8d1f30b 662 rt = rt->dst.rt6_next)
afc154e9 663 match = find_match(rt, oif, strict, &mpri, match, do_rr);
f11e6659 664 for (rt = fn->leaf; rt && rt != rr_head && rt->rt6i_metric == metric;
d8d1f30b 665 rt = rt->dst.rt6_next)
afc154e9 666 match = find_match(rt, oif, strict, &mpri, match, do_rr);
1da177e4 667
f11e6659
DM		 668 return match;
669}
1da177e4 670
f11e6659
DM		 671static struct rt6_info *rt6_select(struct fib6_node *fn, int oif, int strict)
672{
673 struct rt6_info *match, *rt0;
8ed67789 674 struct net *net;
afc154e9 675 bool do_rr = false;
1da177e4 676
f11e6659
DM		 677 rt0 = fn->rr_ptr;
678 if (!rt0)
679 fn->rr_ptr = rt0 = fn->leaf;
1da177e4 680
afc154e9
HFS		 681 match = find_rr_leaf(fn, rt0, rt0->rt6i_metric, oif, strict,
682 &do_rr);
1da177e4 683
afc154e9 684 if (do_rr) {
d8d1f30b 685 struct rt6_info *next = rt0->dst.rt6_next;
f11e6659 686
554cfb7e 687 /* no entries matched; do round-robin */
f11e6659
DM		 688 if (!next || next->rt6i_metric != rt0->rt6i_metric)
689 next = fn->leaf;
690
691 if (next != rt0)
692 fn->rr_ptr = next;
1da177e4 693 }
1da177e4 694
d1918542 695 net = dev_net(rt0->dst.dev);
a02cec21 696 return match ? match : net->ipv6.ip6_null_entry;
1da177e4
LT		 697}
698
70ceb4f5
YH		 699#ifdef CONFIG_IPV6_ROUTE_INFO
700int rt6_route_rcv(struct net_device *dev, u8 *opt, int len,
b71d1d42 701 const struct in6_addr *gwaddr)
70ceb4f5 702{
c346dca1 703 struct net *net = dev_net(dev);
70ceb4f5
YH		 704 struct route_info *rinfo = (struct route_info *) opt;
705 struct in6_addr prefix_buf, *prefix;
706 unsigned int pref;
4bed72e4 707 unsigned long lifetime;
70ceb4f5
YH		 708 struct rt6_info *rt;
709
710 if (len < sizeof(struct route_info)) {
711 return -EINVAL;
712 }
713
714 /* Sanity check for prefix_len and length */
715 if (rinfo->length > 3) {
716 return -EINVAL;
717 } else if (rinfo->prefix_len > 128) {
718 return -EINVAL;
719 } else if (rinfo->prefix_len > 64) {
720 if (rinfo->length < 2) {
721 return -EINVAL;
722 }
723 } else if (rinfo->prefix_len > 0) {
724 if (rinfo->length < 1) {
725 return -EINVAL;
726 }
727 }
728
729 pref = rinfo->route_pref;
730 if (pref == ICMPV6_ROUTER_PREF_INVALID)
3933fc95 731 return -EINVAL;
70ceb4f5 732
4bed72e4 733 lifetime = addrconf_timeout_fixup(ntohl(rinfo->lifetime), HZ);
70ceb4f5
YH		 734
735 if (rinfo->length == 3)
736 prefix = (struct in6_addr *)rinfo->prefix;
737 else {
738 /* this function is safe */
739 ipv6_addr_prefix(&prefix_buf,
740 (struct in6_addr *)rinfo->prefix,
741 rinfo->prefix_len);
742 prefix = &prefix_buf;
743 }
744
f104a567
DJ		 745 if (rinfo->prefix_len == 0)
746 rt = rt6_get_dflt_router(gwaddr, dev);
747 else
748 rt = rt6_get_route_info(net, prefix, rinfo->prefix_len,
749 gwaddr, dev->ifindex);
70ceb4f5
YH		 750
751 if (rt && !lifetime) {
e0a1ad73 752 ip6_del_rt(rt);
70ceb4f5
YH		 753 rt = NULL;
754 }
755
756 if (!rt && lifetime)
efa2cea0 757 rt = rt6_add_route_info(net, prefix, rinfo->prefix_len, gwaddr, dev->ifindex,
70ceb4f5
YH		 758 pref);
759 else if (rt)
760 rt->rt6i_flags = RTF_ROUTEINFO |
761 (rt->rt6i_flags & ~RTF_PREF_MASK) | RTF_PREF(pref);
762
763 if (rt) {
1716a961
G		 764 if (!addrconf_finite_timeout(lifetime))
765 rt6_clean_expires(rt);
766 else
767 rt6_set_expires(rt, jiffies + HZ * lifetime);
768
94e187c0 769 ip6_rt_put(rt);
70ceb4f5
YH		 770 }
771 return 0;
772}
773#endif
774
a3c00e46
MKL		 775static struct fib6_node* fib6_backtrack(struct fib6_node *fn,
776 struct in6_addr *saddr)
777{
778 struct fib6_node *pn;
779 while (1) {
780 if (fn->fn_flags & RTN_TL_ROOT)
781 return NULL;
782 pn = fn->parent;
783 if (FIB6_SUBTREE(pn) && FIB6_SUBTREE(pn) != fn)
784 fn = fib6_lookup(FIB6_SUBTREE(pn), NULL, saddr);
785 else
786 fn = pn;
787 if (fn->fn_flags & RTN_RTINFO)
788 return fn;
789 }
790}
c71099ac 791
8ed67789
DL		 792static struct rt6_info *ip6_pol_route_lookup(struct net *net,
793 struct fib6_table *table,
4c9483b2 794 struct flowi6 *fl6, int flags)
1da177e4
LT		 795{
796 struct fib6_node *fn;
797 struct rt6_info *rt;
798
c71099ac 799 read_lock_bh(&table->tb6_lock);
4c9483b2 800 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
c71099ac
TG		 801restart:
802 rt = fn->leaf;
4c9483b2 803 rt = rt6_device_match(net, rt, &fl6->saddr, fl6->flowi6_oif, flags);
51ebd318 804 if (rt->rt6i_nsiblings && fl6->flowi6_oif == 0)
52bd4c0c 805 rt = rt6_multipath_select(rt, fl6, fl6->flowi6_oif, flags);
a3c00e46
MKL		 806 if (rt == net->ipv6.ip6_null_entry) {
807 fn = fib6_backtrack(fn, &fl6->saddr);
808 if (fn)
809 goto restart;
810 }
d8d1f30b 811 dst_use(&rt->dst, jiffies);
c71099ac 812 read_unlock_bh(&table->tb6_lock);
c71099ac
TG		 813 return rt;
814
815}
816
67ba4152 817struct dst_entry *ip6_route_lookup(struct net *net, struct flowi6 *fl6,
ea6e574e
FW		 818 int flags)
819{
820 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_lookup);
821}
822EXPORT_SYMBOL_GPL(ip6_route_lookup);
823
9acd9f3a
YH		 824struct rt6_info *rt6_lookup(struct net *net, const struct in6_addr *daddr,
825 const struct in6_addr *saddr, int oif, int strict)
c71099ac 826{
4c9483b2
DM		 827 struct flowi6 fl6 = {
828 .flowi6_oif = oif,
829 .daddr = *daddr,
c71099ac
TG		 830 };
831 struct dst_entry *dst;
77d16f45 832 int flags = strict ? RT6_LOOKUP_F_IFACE : 0;
c71099ac 833
adaa70bb 834 if (saddr) {
4c9483b2 835 memcpy(&fl6.saddr, saddr, sizeof(*saddr));
adaa70bb
TG		 836 flags |= RT6_LOOKUP_F_HAS_SADDR;
837 }
838
4c9483b2 839 dst = fib6_rule_lookup(net, &fl6, flags, ip6_pol_route_lookup);
c71099ac
TG		 840 if (dst->error == 0)
841 return (struct rt6_info *) dst;
842
843 dst_release(dst);
844
1da177e4
LT		 845 return NULL;
846}
7159039a
YH		 847EXPORT_SYMBOL(rt6_lookup);
848
c71099ac 849/* ip6_ins_rt is called with FREE table->tb6_lock.
1da177e4
LT		 850 It takes new route entry, the addition fails by any reason the
851 route is freed. In any case, if caller does not hold it, it may
852 be destroyed.
853 */
854
e5fd387a 855static int __ip6_ins_rt(struct rt6_info *rt, struct nl_info *info,
e715b6d3 856 struct mx6_config *mxc)
1da177e4
LT		 857{
858 int err;
c71099ac 859 struct fib6_table *table;
1da177e4 860
c71099ac
TG		 861 table = rt->rt6i_table;
862 write_lock_bh(&table->tb6_lock);
e715b6d3 863 err = fib6_add(&table->tb6_root, rt, info, mxc);
c71099ac 864 write_unlock_bh(&table->tb6_lock);
1da177e4
LT		 865
866 return err;
867}
868
40e22e8f
TG		 869int ip6_ins_rt(struct rt6_info *rt)
870{
e715b6d3
FW		 871 struct nl_info info = { .nl_net = dev_net(rt->dst.dev), };
872 struct mx6_config mxc = { .mx = NULL, };
873
874 return __ip6_ins_rt(rt, &info, &mxc);
40e22e8f
TG		 875}
876
1716a961 877static struct rt6_info *rt6_alloc_cow(struct rt6_info *ort,
21efcfa0 878 const struct in6_addr *daddr,
b71d1d42 879 const struct in6_addr *saddr)
1da177e4 880{
1da177e4
LT		 881 struct rt6_info *rt;
882
883 /*
884 * Clone the route.
885 */
886
21efcfa0 887 rt = ip6_rt_copy(ort, daddr);
1da177e4
LT		 888
889 if (rt) {
249a3630
DJ		 890 if (ort->rt6i_dst.plen != 128 &&
891 ipv6_addr_equal(&ort->rt6i_dst.addr, daddr))
892 rt->rt6i_flags |= RTF_ANYCAST;
1da177e4 893
1da177e4 894 rt->rt6i_flags |= RTF_CACHE;
1da177e4
LT		 895
896#ifdef CONFIG_IPV6_SUBTREES
897 if (rt->rt6i_src.plen && saddr) {
4e3fd7a0 898 rt->rt6i_src.addr = *saddr;
1da177e4
LT		 899 rt->rt6i_src.plen = 128;
900 }
901#endif
95a9a5ba 902 }
1da177e4 903
95a9a5ba
YH		 904 return rt;
905}
1da177e4 906
21efcfa0
ED		 907static struct rt6_info *rt6_alloc_clone(struct rt6_info *ort,
908 const struct in6_addr *daddr)
299d9939 909{
21efcfa0
ED		 910 struct rt6_info *rt = ip6_rt_copy(ort, daddr);
911
887c95cc 912 if (rt)
299d9939 913 rt->rt6i_flags |= RTF_CACHE;
299d9939
YH		 914 return rt;
915}
916
8ed67789 917static struct rt6_info *ip6_pol_route(struct net *net, struct fib6_table *table, int oif,
4c9483b2 918 struct flowi6 *fl6, int flags)
1da177e4 919{
367efcb9 920 struct fib6_node *fn, *saved_fn;
519fbd87 921 struct rt6_info *rt, *nrt;
c71099ac 922 int strict = 0;
1da177e4 923 int attempts = 3;
519fbd87 924 int err;
1da177e4 925
77d16f45 926 strict |= flags & RT6_LOOKUP_F_IFACE;
367efcb9
MKL		 927 if (net->ipv6.devconf_all->forwarding == 0)
928 strict |= RT6_LOOKUP_F_REACHABLE;
1da177e4 929
a3c00e46 930redo_fib6_lookup_lock:
c71099ac 931 read_lock_bh(&table->tb6_lock);
1da177e4 932
4c9483b2 933 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
367efcb9 934 saved_fn = fn;
1da177e4 935
a3c00e46 936redo_rt6_select:
367efcb9 937 rt = rt6_select(fn, oif, strict);
52bd4c0c 938 if (rt->rt6i_nsiblings)
367efcb9 939 rt = rt6_multipath_select(rt, fl6, oif, strict);
a3c00e46
MKL		 940 if (rt == net->ipv6.ip6_null_entry) {
941 fn = fib6_backtrack(fn, &fl6->saddr);
942 if (fn)
943 goto redo_rt6_select;
367efcb9
MKL		 944 else if (strict & RT6_LOOKUP_F_REACHABLE) {
945 /* also consider unreachable route */
946 strict &= ~RT6_LOOKUP_F_REACHABLE;
947 fn = saved_fn;
948 goto redo_rt6_select;
949 } else {
950 dst_hold(&rt->dst);
951 read_unlock_bh(&table->tb6_lock);
952 goto out2;
953 }
a3c00e46
MKL		 954 }
955
d8d1f30b 956 dst_hold(&rt->dst);
c71099ac 957 read_unlock_bh(&table->tb6_lock);
fb9de91e 958
94c77bb4
MKL		 959 if (rt->rt6i_flags & RTF_CACHE)
960 goto out2;
961
c440f160 962 if (!(rt->rt6i_flags & (RTF_NONEXTHOP | RTF_GATEWAY)))
4c9483b2 963 nrt = rt6_alloc_cow(rt, &fl6->daddr, &fl6->saddr);
7343ff31 964 else if (!(rt->dst.flags & DST_HOST))
4c9483b2 965 nrt = rt6_alloc_clone(rt, &fl6->daddr);
7343ff31
DM		 966 else
967 goto out2;
e40cf353 968
94e187c0 969 ip6_rt_put(rt);
8ed67789 970 rt = nrt ? : net->ipv6.ip6_null_entry;
1da177e4 971
d8d1f30b 972 dst_hold(&rt->dst);
519fbd87 973 if (nrt) {
40e22e8f 974 err = ip6_ins_rt(nrt);
519fbd87 975 if (!err)
1da177e4 976 goto out2;
1da177e4 977 }
1da177e4 978
519fbd87
YH		 979 if (--attempts <= 0)
980 goto out2;
981
982 /*
c71099ac 983 * Race condition! In the gap, when table->tb6_lock was
519fbd87
YH		 984 * released someone could insert this route. Relookup.
985 */
94e187c0 986 ip6_rt_put(rt);
a3c00e46 987 goto redo_fib6_lookup_lock;
519fbd87 988
1da177e4 989out2:
d8d1f30b
CG		 990 rt->dst.lastuse = jiffies;
991 rt->dst.__use++;
c71099ac
TG		 992
993 return rt;
1da177e4
LT		 994}
995
8ed67789 996static struct rt6_info *ip6_pol_route_input(struct net *net, struct fib6_table *table,
4c9483b2 997 struct flowi6 *fl6, int flags)
4acad72d 998{
4c9483b2 999 return ip6_pol_route(net, table, fl6->flowi6_iif, fl6, flags);
4acad72d
PE		 1000}
1001
72331bc0
SL		 1002static struct dst_entry *ip6_route_input_lookup(struct net *net,
1003 struct net_device *dev,
1004 struct flowi6 *fl6, int flags)
1005{
1006 if (rt6_need_strict(&fl6->daddr) && dev->type != ARPHRD_PIMREG)
1007 flags |= RT6_LOOKUP_F_IFACE;
1008
1009 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_input);
1010}
1011
c71099ac
TG		 1012void ip6_route_input(struct sk_buff *skb)
1013{
b71d1d42 1014 const struct ipv6hdr *iph = ipv6_hdr(skb);
c346dca1 1015 struct net *net = dev_net(skb->dev);
adaa70bb 1016 int flags = RT6_LOOKUP_F_HAS_SADDR;
4c9483b2
DM		 1017 struct flowi6 fl6 = {
1018 .flowi6_iif = skb->dev->ifindex,
1019 .daddr = iph->daddr,
1020 .saddr = iph->saddr,
6502ca52 1021 .flowlabel = ip6_flowinfo(iph),
4c9483b2
DM		 1022 .flowi6_mark = skb->mark,
1023 .flowi6_proto = iph->nexthdr,
c71099ac 1024 };
adaa70bb 1025
72331bc0 1026 skb_dst_set(skb, ip6_route_input_lookup(net, skb->dev, &fl6, flags));
c71099ac
TG		 1027}
1028
8ed67789 1029static struct rt6_info *ip6_pol_route_output(struct net *net, struct fib6_table *table,
4c9483b2 1030 struct flowi6 *fl6, int flags)
1da177e4 1031{
4c9483b2 1032 return ip6_pol_route(net, table, fl6->flowi6_oif, fl6, flags);
c71099ac
TG		 1033}
1034
67ba4152 1035struct dst_entry *ip6_route_output(struct net *net, const struct sock *sk,
4c9483b2 1036 struct flowi6 *fl6)
c71099ac
TG		 1037{
1038 int flags = 0;
1039
1fb9489b 1040 fl6->flowi6_iif = LOOPBACK_IFINDEX;
4dc27d1c 1041
4c9483b2 1042 if ((sk && sk->sk_bound_dev_if) || rt6_need_strict(&fl6->daddr))
77d16f45 1043 flags |= RT6_LOOKUP_F_IFACE;
c71099ac 1044
4c9483b2 1045 if (!ipv6_addr_any(&fl6->saddr))
adaa70bb 1046 flags |= RT6_LOOKUP_F_HAS_SADDR;
0c9a2ac1
YH		 1047 else if (sk)
1048 flags |= rt6_srcprefs2flags(inet6_sk(sk)->srcprefs);
adaa70bb 1049
4c9483b2 1050 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_output);
1da177e4 1051}
7159039a 1052EXPORT_SYMBOL(ip6_route_output);
1da177e4 1053
2774c131 1054struct dst_entry *ip6_blackhole_route(struct net *net, struct dst_entry *dst_orig)
14e50e57 1055{
5c1e6aa3 1056 struct rt6_info *rt, *ort = (struct rt6_info *) dst_orig;
14e50e57
DM		 1057 struct dst_entry *new = NULL;
1058
f5b0a874 1059 rt = dst_alloc(&ip6_dst_blackhole_ops, ort->dst.dev, 1, DST_OBSOLETE_NONE, 0);
14e50e57 1060 if (rt) {
d8d1f30b 1061 new = &rt->dst;
14e50e57 1062
8104891b
SK		 1063 memset(new + 1, 0, sizeof(*rt) - sizeof(*new));
1064 rt6_init_peer(rt, net->ipv6.peers);
1065
14e50e57 1066 new->__use = 1;
352e512c 1067 new->input = dst_discard;
aad88724 1068 new->output = dst_discard_sk;
14e50e57 1069
21efcfa0
ED		 1070 if (dst_metrics_read_only(&ort->dst))
1071 new->_metrics = ort->dst._metrics;
1072 else
1073 dst_copy_metrics(new, &ort->dst);
14e50e57
DM		 1074 rt->rt6i_idev = ort->rt6i_idev;
1075 if (rt->rt6i_idev)
1076 in6_dev_hold(rt->rt6i_idev);
14e50e57 1077
4e3fd7a0 1078 rt->rt6i_gateway = ort->rt6i_gateway;
1716a961 1079 rt->rt6i_flags = ort->rt6i_flags;
14e50e57
DM		 1080 rt->rt6i_metric = 0;
1081
1082 memcpy(&rt->rt6i_dst, &ort->rt6i_dst, sizeof(struct rt6key));
1083#ifdef CONFIG_IPV6_SUBTREES
1084 memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key));
1085#endif
1086
1087 dst_free(new);
1088 }
1089
69ead7af
DM		 1090 dst_release(dst_orig);
1091 return new ? new : ERR_PTR(-ENOMEM);
14e50e57 1092}
14e50e57 1093
1da177e4
LT		 1094/*
1095 * Destination cache support functions
1096 */
1097
1098static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie)
1099{
1100 struct rt6_info *rt;
1101
1102 rt = (struct rt6_info *) dst;
1103
6f3118b5
ND		 1104 /* All IPV6 dsts are created with ->obsolete set to the value
1105 * DST_OBSOLETE_FORCE_CHK which forces validation calls down
1106 * into this function always.
1107 */
e3bc10bd
HFS		 1108 if (!rt->rt6i_node || (rt->rt6i_node->fn_sernum != cookie))
1109 return NULL;
a4477c4d 1110
e3bc10bd
HFS		 1111 if (rt6_check_expired(rt))
1112 return NULL;
1113
1114 return dst;
1da177e4
LT		 1115}
1116
1117static struct dst_entry *ip6_negative_advice(struct dst_entry *dst)
1118{
1119 struct rt6_info *rt = (struct rt6_info *) dst;
1120
1121 if (rt) {
54c1a859
YH		 1122 if (rt->rt6i_flags & RTF_CACHE) {
1123 if (rt6_check_expired(rt)) {
1124 ip6_del_rt(rt);
1125 dst = NULL;
1126 }
1127 } else {
1da177e4 1128 dst_release(dst);
54c1a859
YH		 1129 dst = NULL;
1130 }
1da177e4 1131 }
54c1a859 1132 return dst;
1da177e4
LT		 1133}
1134
1135static void ip6_link_failure(struct sk_buff *skb)
1136{
1137 struct rt6_info *rt;
1138
3ffe533c 1139 icmpv6_send(skb, ICMPV6_DEST_UNREACH, ICMPV6_ADDR_UNREACH, 0);
1da177e4 1140
adf30907 1141 rt = (struct rt6_info *) skb_dst(skb);
1da177e4 1142 if (rt) {
1eb4f758
HFS		 1143 if (rt->rt6i_flags & RTF_CACHE) {
1144 dst_hold(&rt->dst);
1145 if (ip6_del_rt(rt))
1146 dst_free(&rt->dst);
1147 } else if (rt->rt6i_node && (rt->rt6i_flags & RTF_DEFAULT)) {
1da177e4 1148 rt->rt6i_node->fn_sernum = -1;
1eb4f758 1149 }
1da177e4
LT		 1150 }
1151}
1152
6700c270
DM		 1153static void ip6_rt_update_pmtu(struct dst_entry *dst, struct sock *sk,
1154 struct sk_buff *skb, u32 mtu)
1da177e4 1155{
67ba4152 1156 struct rt6_info *rt6 = (struct rt6_info *)dst;
1da177e4 1157
81aded24 1158 dst_confirm(dst);
1da177e4 1159 if (mtu < dst_mtu(dst) && rt6->rt6i_dst.plen == 128) {
81aded24
DM		 1160 struct net *net = dev_net(dst->dev);
1161
1da177e4 1162 rt6->rt6i_flags |= RTF_MODIFIED;
9d289715 1163 if (mtu < IPV6_MIN_MTU)
1da177e4 1164 mtu = IPV6_MIN_MTU;
9d289715 1165
defb3519 1166 dst_metric_set(dst, RTAX_MTU, mtu);
81aded24 1167 rt6_update_expires(rt6, net->ipv6.sysctl.ip6_rt_mtu_expires);
1da177e4
LT		 1168 }
1169}
1170
42ae66c8
DM		 1171void ip6_update_pmtu(struct sk_buff *skb, struct net *net, __be32 mtu,
1172 int oif, u32 mark)
81aded24
DM		 1173{
1174 const struct ipv6hdr *iph = (struct ipv6hdr *) skb->data;
1175 struct dst_entry *dst;
1176 struct flowi6 fl6;
1177
1178 memset(&fl6, 0, sizeof(fl6));
1179 fl6.flowi6_oif = oif;
1b3c61dc 1180 fl6.flowi6_mark = mark ? mark : IP6_REPLY_MARK(net, skb->mark);
81aded24
DM		 1181 fl6.daddr = iph->daddr;
1182 fl6.saddr = iph->saddr;
6502ca52 1183 fl6.flowlabel = ip6_flowinfo(iph);
81aded24
DM		 1184
1185 dst = ip6_route_output(net, NULL, &fl6);
1186 if (!dst->error)
6700c270 1187 ip6_rt_update_pmtu(dst, NULL, skb, ntohl(mtu));
81aded24
DM		 1188 dst_release(dst);
1189}
1190EXPORT_SYMBOL_GPL(ip6_update_pmtu);
1191
1192void ip6_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, __be32 mtu)
1193{
1194 ip6_update_pmtu(skb, sock_net(sk), mtu,
1195 sk->sk_bound_dev_if, sk->sk_mark);
1196}
1197EXPORT_SYMBOL_GPL(ip6_sk_update_pmtu);
1198
b55b76b2
DJ		 1199/* Handle redirects */
1200struct ip6rd_flowi {
1201 struct flowi6 fl6;
1202 struct in6_addr gateway;
1203};
1204
1205static struct rt6_info *__ip6_route_redirect(struct net *net,
1206 struct fib6_table *table,
1207 struct flowi6 *fl6,
1208 int flags)
1209{
1210 struct ip6rd_flowi *rdfl = (struct ip6rd_flowi *)fl6;
1211 struct rt6_info *rt;
1212 struct fib6_node *fn;
1213
1214 /* Get the "current" route for this destination and
1215 * check if the redirect has come from approriate router.
1216 *
1217 * RFC 4861 specifies that redirects should only be
1218 * accepted if they come from the nexthop to the target.
1219 * Due to the way the routes are chosen, this notion
1220 * is a bit fuzzy and one might need to check all possible
1221 * routes.
1222 */
1223
1224 read_lock_bh(&table->tb6_lock);
1225 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
1226restart:
1227 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
1228 if (rt6_check_expired(rt))
1229 continue;
1230 if (rt->dst.error)
1231 break;
1232 if (!(rt->rt6i_flags & RTF_GATEWAY))
1233 continue;
1234 if (fl6->flowi6_oif != rt->dst.dev->ifindex)
1235 continue;
1236 if (!ipv6_addr_equal(&rdfl->gateway, &rt->rt6i_gateway))
1237 continue;
1238 break;
1239 }
1240
1241 if (!rt)
1242 rt = net->ipv6.ip6_null_entry;
1243 else if (rt->dst.error) {
1244 rt = net->ipv6.ip6_null_entry;
b0a1ba59
MKL		 1245 goto out;
1246 }
1247
1248 if (rt == net->ipv6.ip6_null_entry) {
a3c00e46
MKL		 1249 fn = fib6_backtrack(fn, &fl6->saddr);
1250 if (fn)
1251 goto restart;
b55b76b2 1252 }
a3c00e46 1253
b0a1ba59 1254out:
b55b76b2
DJ		 1255 dst_hold(&rt->dst);
1256
1257 read_unlock_bh(&table->tb6_lock);
1258
1259 return rt;
1260};
1261
1262static struct dst_entry *ip6_route_redirect(struct net *net,
1263 const struct flowi6 *fl6,
1264 const struct in6_addr *gateway)
1265{
1266 int flags = RT6_LOOKUP_F_HAS_SADDR;
1267 struct ip6rd_flowi rdfl;
1268
1269 rdfl.fl6 = *fl6;
1270 rdfl.gateway = *gateway;
1271
1272 return fib6_rule_lookup(net, &rdfl.fl6,
1273 flags, __ip6_route_redirect);
1274}
1275
3a5ad2ee
DM		 1276void ip6_redirect(struct sk_buff *skb, struct net *net, int oif, u32 mark)
1277{
1278 const struct ipv6hdr *iph = (struct ipv6hdr *) skb->data;
1279 struct dst_entry *dst;
1280 struct flowi6 fl6;
1281
1282 memset(&fl6, 0, sizeof(fl6));
e374c618 1283 fl6.flowi6_iif = LOOPBACK_IFINDEX;
3a5ad2ee
DM		 1284 fl6.flowi6_oif = oif;
1285 fl6.flowi6_mark = mark;
3a5ad2ee
DM		 1286 fl6.daddr = iph->daddr;
1287 fl6.saddr = iph->saddr;
6502ca52 1288 fl6.flowlabel = ip6_flowinfo(iph);
3a5ad2ee 1289
b55b76b2
DJ		 1290 dst = ip6_route_redirect(net, &fl6, &ipv6_hdr(skb)->saddr);
1291 rt6_do_redirect(dst, NULL, skb);
3a5ad2ee
DM		 1292 dst_release(dst);
1293}
1294EXPORT_SYMBOL_GPL(ip6_redirect);
1295
c92a59ec
DJ		 1296void ip6_redirect_no_header(struct sk_buff *skb, struct net *net, int oif,
1297 u32 mark)
1298{
1299 const struct ipv6hdr *iph = ipv6_hdr(skb);
1300 const struct rd_msg *msg = (struct rd_msg *)icmp6_hdr(skb);
1301 struct dst_entry *dst;
1302 struct flowi6 fl6;
1303
1304 memset(&fl6, 0, sizeof(fl6));
e374c618 1305 fl6.flowi6_iif = LOOPBACK_IFINDEX;
c92a59ec
DJ		 1306 fl6.flowi6_oif = oif;
1307 fl6.flowi6_mark = mark;
c92a59ec
DJ		 1308 fl6.daddr = msg->dest;
1309 fl6.saddr = iph->daddr;
1310
b55b76b2
DJ		 1311 dst = ip6_route_redirect(net, &fl6, &iph->saddr);
1312 rt6_do_redirect(dst, NULL, skb);
c92a59ec
DJ		 1313 dst_release(dst);
1314}
1315
3a5ad2ee
DM		 1316void ip6_sk_redirect(struct sk_buff *skb, struct sock *sk)
1317{
1318 ip6_redirect(skb, sock_net(sk), sk->sk_bound_dev_if, sk->sk_mark);
1319}
1320EXPORT_SYMBOL_GPL(ip6_sk_redirect);
1321
0dbaee3b 1322static unsigned int ip6_default_advmss(const struct dst_entry *dst)
1da177e4 1323{
0dbaee3b
DM		 1324 struct net_device *dev = dst->dev;
1325 unsigned int mtu = dst_mtu(dst);
1326 struct net *net = dev_net(dev);
1327
1da177e4
LT		 1328 mtu -= sizeof(struct ipv6hdr) + sizeof(struct tcphdr);
1329
5578689a
DL		 1330 if (mtu < net->ipv6.sysctl.ip6_rt_min_advmss)
1331 mtu = net->ipv6.sysctl.ip6_rt_min_advmss;
1da177e4
LT		 1332
1333 /*
1ab1457c
YH		 1334 * Maximal non-jumbo IPv6 payload is IPV6_MAXPLEN and
1335 * corresponding MSS is IPV6_MAXPLEN - tcp_header_size.
1336 * IPV6_MAXPLEN is also valid and means: "any MSS,
1da177e4
LT		 1337 * rely only on pmtu discovery"
1338 */
1339 if (mtu > IPV6_MAXPLEN - sizeof(struct tcphdr))
1340 mtu = IPV6_MAXPLEN;
1341 return mtu;
1342}
1343
ebb762f2 1344static unsigned int ip6_mtu(const struct dst_entry *dst)
d33e4553 1345{
d33e4553 1346 struct inet6_dev *idev;
618f9bc7
SK		 1347 unsigned int mtu = dst_metric_raw(dst, RTAX_MTU);
1348
1349 if (mtu)
30f78d8e 1350 goto out;
618f9bc7
SK		 1351
1352 mtu = IPV6_MIN_MTU;
d33e4553
DM		 1353
1354 rcu_read_lock();
1355 idev = __in6_dev_get(dst->dev);
1356 if (idev)
1357 mtu = idev->cnf.mtu6;
1358 rcu_read_unlock();
1359
30f78d8e
ED		 1360out:
1361 return min_t(unsigned int, mtu, IP6_MAX_MTU);
d33e4553
DM		 1362}
1363
3b00944c
YH		 1364static struct dst_entry *icmp6_dst_gc_list;
1365static DEFINE_SPINLOCK(icmp6_dst_lock);
5d0bbeeb 1366
3b00944c 1367struct dst_entry *icmp6_dst_alloc(struct net_device *dev,
87a11578 1368 struct flowi6 *fl6)
1da177e4 1369{
87a11578 1370 struct dst_entry *dst;
1da177e4
LT		 1371 struct rt6_info *rt;
1372 struct inet6_dev *idev = in6_dev_get(dev);
c346dca1 1373 struct net *net = dev_net(dev);
1da177e4 1374
38308473 1375 if (unlikely(!idev))
122bdf67 1376 return ERR_PTR(-ENODEV);
1da177e4 1377
8b96d22d 1378 rt = ip6_dst_alloc(net, dev, 0, NULL);
38308473 1379 if (unlikely(!rt)) {
1da177e4 1380 in6_dev_put(idev);
87a11578 1381 dst = ERR_PTR(-ENOMEM);
1da177e4
LT		 1382 goto out;
1383 }
1384
8e2ec639
YZ		 1385 rt->dst.flags |= DST_HOST;
1386 rt->dst.output = ip6_output;
d8d1f30b 1387 atomic_set(&rt->dst.__refcnt, 1);
550bab42 1388 rt->rt6i_gateway = fl6->daddr;
87a11578 1389 rt->rt6i_dst.addr = fl6->daddr;
8e2ec639
YZ		 1390 rt->rt6i_dst.plen = 128;
1391 rt->rt6i_idev = idev;
14edd87d 1392 dst_metric_set(&rt->dst, RTAX_HOPLIMIT, 0);
1da177e4 1393
3b00944c 1394 spin_lock_bh(&icmp6_dst_lock);
d8d1f30b
CG		 1395 rt->dst.next = icmp6_dst_gc_list;
1396 icmp6_dst_gc_list = &rt->dst;
3b00944c 1397 spin_unlock_bh(&icmp6_dst_lock);
1da177e4 1398
5578689a 1399 fib6_force_start_gc(net);
1da177e4 1400
87a11578
DM		 1401 dst = xfrm_lookup(net, &rt->dst, flowi6_to_flowi(fl6), NULL, 0);
1402
1da177e4 1403out:
87a11578 1404 return dst;
1da177e4
LT		 1405}
1406
3d0f24a7 1407int icmp6_dst_gc(void)
1da177e4 1408{
e9476e95 1409 struct dst_entry *dst, **pprev;
3d0f24a7 1410 int more = 0;
1da177e4 1411
3b00944c
YH		 1412 spin_lock_bh(&icmp6_dst_lock);
1413 pprev = &icmp6_dst_gc_list;
5d0bbeeb 1414
1da177e4
LT		 1415 while ((dst = *pprev) != NULL) {
1416 if (!atomic_read(&dst->__refcnt)) {
1417 *pprev = dst->next;
1418 dst_free(dst);
1da177e4
LT		 1419 } else {
1420 pprev = &dst->next;
3d0f24a7 1421 ++more;
1da177e4
LT		 1422 }
1423 }
1424
3b00944c 1425 spin_unlock_bh(&icmp6_dst_lock);
5d0bbeeb 1426
3d0f24a7 1427 return more;
1da177e4
LT		 1428}
1429
1e493d19
DM		 1430static void icmp6_clean_all(int (*func)(struct rt6_info *rt, void *arg),
1431 void *arg)
1432{
1433 struct dst_entry *dst, **pprev;
1434
1435 spin_lock_bh(&icmp6_dst_lock);
1436 pprev = &icmp6_dst_gc_list;
1437 while ((dst = *pprev) != NULL) {
1438 struct rt6_info *rt = (struct rt6_info *) dst;
1439 if (func(rt, arg)) {
1440 *pprev = dst->next;
1441 dst_free(dst);
1442 } else {
1443 pprev = &dst->next;
1444 }
1445 }
1446 spin_unlock_bh(&icmp6_dst_lock);
1447}
1448
569d3645 1449static int ip6_dst_gc(struct dst_ops *ops)
1da177e4 1450{
86393e52 1451 struct net *net = container_of(ops, struct net, ipv6.ip6_dst_ops);
7019b78e
DL		 1452 int rt_min_interval = net->ipv6.sysctl.ip6_rt_gc_min_interval;
1453 int rt_max_size = net->ipv6.sysctl.ip6_rt_max_size;
1454 int rt_elasticity = net->ipv6.sysctl.ip6_rt_gc_elasticity;
1455 int rt_gc_timeout = net->ipv6.sysctl.ip6_rt_gc_timeout;
1456 unsigned long rt_last_gc = net->ipv6.ip6_rt_last_gc;
fc66f95c 1457 int entries;
7019b78e 1458
fc66f95c 1459 entries = dst_entries_get_fast(ops);
49a18d86 1460 if (time_after(rt_last_gc + rt_min_interval, jiffies) &&
fc66f95c 1461 entries <= rt_max_size)
1da177e4
LT		 1462 goto out;
1463
6891a346 1464 net->ipv6.ip6_rt_gc_expire++;
14956643 1465 fib6_run_gc(net->ipv6.ip6_rt_gc_expire, net, true);
fc66f95c
ED		 1466 entries = dst_entries_get_slow(ops);
1467 if (entries < ops->gc_thresh)
7019b78e 1468 net->ipv6.ip6_rt_gc_expire = rt_gc_timeout>>1;
1da177e4 1469out:
7019b78e 1470 net->ipv6.ip6_rt_gc_expire -= net->ipv6.ip6_rt_gc_expire>>rt_elasticity;
fc66f95c 1471 return entries > rt_max_size;
1da177e4
LT		 1472}
1473
e715b6d3
FW		 1474static int ip6_convert_metrics(struct mx6_config *mxc,
1475 const struct fib6_config *cfg)
1476{
1477 struct nlattr *nla;
1478 int remaining;
1479 u32 *mp;
1480
1481 if (cfg->fc_mx == NULL)
1482 return 0;
1483
1484 mp = kzalloc(sizeof(u32) * RTAX_MAX, GFP_KERNEL);
1485 if (unlikely(!mp))
1486 return -ENOMEM;
1487
1488 nla_for_each_attr(nla, cfg->fc_mx, cfg->fc_mx_len, remaining) {
1489 int type = nla_type(nla);
1490
1491 if (type) {
ea697639
DB		 1492 u32 val;
1493
e715b6d3
FW		 1494 if (unlikely(type > RTAX_MAX))
1495 goto err;
ea697639
DB		 1496 if (type == RTAX_CC_ALGO) {
1497 char tmp[TCP_CA_NAME_MAX];
1498
1499 nla_strlcpy(tmp, nla, sizeof(tmp));
1500 val = tcp_ca_get_key_by_name(tmp);
1501 if (val == TCP_CA_UNSPEC)
1502 goto err;
1503 } else {
1504 val = nla_get_u32(nla);
1505 }
e715b6d3 1506
ea697639 1507 mp[type - 1] = val;
e715b6d3
FW		 1508 __set_bit(type - 1, mxc->mx_valid);
1509 }
1510 }
1511
1512 mxc->mx = mp;
1513
1514 return 0;
1515 err:
1516 kfree(mp);
1517 return -EINVAL;
1518}
1da177e4 1519
86872cb5 1520int ip6_route_add(struct fib6_config *cfg)
1da177e4
LT		 1521{
1522 int err;
5578689a 1523 struct net *net = cfg->fc_nlinfo.nl_net;
1da177e4
LT		 1524 struct rt6_info *rt = NULL;
1525 struct net_device *dev = NULL;
1526 struct inet6_dev *idev = NULL;
c71099ac 1527 struct fib6_table *table;
e715b6d3 1528 struct mx6_config mxc = { .mx = NULL, };
1da177e4
LT		 1529 int addr_type;
1530
86872cb5 1531 if (cfg->fc_dst_len > 128 || cfg->fc_src_len > 128)
1da177e4
LT		 1532 return -EINVAL;
1533#ifndef CONFIG_IPV6_SUBTREES
86872cb5 1534 if (cfg->fc_src_len)
1da177e4
LT		 1535 return -EINVAL;
1536#endif
86872cb5 1537 if (cfg->fc_ifindex) {
1da177e4 1538 err = -ENODEV;
5578689a 1539 dev = dev_get_by_index(net, cfg->fc_ifindex);
1da177e4
LT		 1540 if (!dev)
1541 goto out;
1542 idev = in6_dev_get(dev);
1543 if (!idev)
1544 goto out;
1545 }
1546
86872cb5
TG		 1547 if (cfg->fc_metric == 0)
1548 cfg->fc_metric = IP6_RT_PRIO_USER;
1da177e4 1549
d71314b4 1550 err = -ENOBUFS;
38308473
DM		 1551 if (cfg->fc_nlinfo.nlh &&
1552 !(cfg->fc_nlinfo.nlh->nlmsg_flags & NLM_F_CREATE)) {
d71314b4 1553 table = fib6_get_table(net, cfg->fc_table);
38308473 1554 if (!table) {
f3213831 1555 pr_warn("NLM_F_CREATE should be specified when creating new route\n");
d71314b4
MV		 1556 table = fib6_new_table(net, cfg->fc_table);
1557 }
1558 } else {
1559 table = fib6_new_table(net, cfg->fc_table);
1560 }
38308473
DM		 1561
1562 if (!table)
c71099ac 1563 goto out;
c71099ac 1564
c88507fb 1565 rt = ip6_dst_alloc(net, NULL, (cfg->fc_flags & RTF_ADDRCONF) ? 0 : DST_NOCOUNT, table);
1da177e4 1566
38308473 1567 if (!rt) {
1da177e4
LT		 1568 err = -ENOMEM;
1569 goto out;
1570 }
1571
1716a961
G		 1572 if (cfg->fc_flags & RTF_EXPIRES)
1573 rt6_set_expires(rt, jiffies +
1574 clock_t_to_jiffies(cfg->fc_expires));
1575 else
1576 rt6_clean_expires(rt);
1da177e4 1577
86872cb5
TG		 1578 if (cfg->fc_protocol == RTPROT_UNSPEC)
1579 cfg->fc_protocol = RTPROT_BOOT;
1580 rt->rt6i_protocol = cfg->fc_protocol;
1581
1582 addr_type = ipv6_addr_type(&cfg->fc_dst);
1da177e4
LT		 1583
1584 if (addr_type & IPV6_ADDR_MULTICAST)
d8d1f30b 1585 rt->dst.input = ip6_mc_input;
ab79ad14
MŻ		 1586 else if (cfg->fc_flags & RTF_LOCAL)
1587 rt->dst.input = ip6_input;
1da177e4 1588 else
d8d1f30b 1589 rt->dst.input = ip6_forward;
1da177e4 1590
d8d1f30b 1591 rt->dst.output = ip6_output;
1da177e4 1592
86872cb5
TG		 1593 ipv6_addr_prefix(&rt->rt6i_dst.addr, &cfg->fc_dst, cfg->fc_dst_len);
1594 rt->rt6i_dst.plen = cfg->fc_dst_len;
e5fd387a
MK		 1595 if (rt->rt6i_dst.plen == 128) {
1596 rt->dst.flags |= DST_HOST;
1597 dst_metrics_set_force_overwrite(&rt->dst);
8e2ec639 1598 }
e5fd387a 1599
1da177e4 1600#ifdef CONFIG_IPV6_SUBTREES
86872cb5
TG		 1601 ipv6_addr_prefix(&rt->rt6i_src.addr, &cfg->fc_src, cfg->fc_src_len);
1602 rt->rt6i_src.plen = cfg->fc_src_len;
1da177e4
LT		 1603#endif
1604
86872cb5 1605 rt->rt6i_metric = cfg->fc_metric;
1da177e4
LT		 1606
1607 /* We cannot add true routes via loopback here,
1608 they would result in kernel looping; promote them to reject routes
1609 */
86872cb5 1610 if ((cfg->fc_flags & RTF_REJECT) ||
38308473
DM		 1611 (dev && (dev->flags & IFF_LOOPBACK) &&
1612 !(addr_type & IPV6_ADDR_LOOPBACK) &&
1613 !(cfg->fc_flags & RTF_LOCAL))) {
1da177e4 1614 /* hold loopback dev/idev if we haven't done so. */
5578689a 1615 if (dev != net->loopback_dev) {
1da177e4
LT		 1616 if (dev) {
1617 dev_put(dev);
1618 in6_dev_put(idev);
1619 }
5578689a 1620 dev = net->loopback_dev;
1da177e4
LT		 1621 dev_hold(dev);
1622 idev = in6_dev_get(dev);
1623 if (!idev) {
1624 err = -ENODEV;
1625 goto out;
1626 }
1627 }
1da177e4 1628 rt->rt6i_flags = RTF_REJECT|RTF_NONEXTHOP;
ef2c7d7b
ND		 1629 switch (cfg->fc_type) {
1630 case RTN_BLACKHOLE:
1631 rt->dst.error = -EINVAL;
aad88724 1632 rt->dst.output = dst_discard_sk;
7150aede 1633 rt->dst.input = dst_discard;
ef2c7d7b
ND		 1634 break;
1635 case RTN_PROHIBIT:
1636 rt->dst.error = -EACCES;
7150aede
K		 1637 rt->dst.output = ip6_pkt_prohibit_out;
1638 rt->dst.input = ip6_pkt_prohibit;
ef2c7d7b 1639 break;
b4949ab2 1640 case RTN_THROW:
ef2c7d7b 1641 default:
7150aede
K		 1642 rt->dst.error = (cfg->fc_type == RTN_THROW) ? -EAGAIN
1643 : -ENETUNREACH;
1644 rt->dst.output = ip6_pkt_discard_out;
1645 rt->dst.input = ip6_pkt_discard;
ef2c7d7b
ND		 1646 break;
1647 }
1da177e4
LT		 1648 goto install_route;
1649 }
1650
86872cb5 1651 if (cfg->fc_flags & RTF_GATEWAY) {
b71d1d42 1652 const struct in6_addr *gw_addr;
1da177e4
LT		 1653 int gwa_type;
1654
86872cb5 1655 gw_addr = &cfg->fc_gateway;
4e3fd7a0 1656 rt->rt6i_gateway = *gw_addr;
1da177e4
LT		 1657 gwa_type = ipv6_addr_type(gw_addr);
1658
1659 if (gwa_type != (IPV6_ADDR_LINKLOCAL|IPV6_ADDR_UNICAST)) {
1660 struct rt6_info *grt;
1661
1662 /* IPv6 strictly inhibits using not link-local
1663 addresses as nexthop address.
1664 Otherwise, router will not able to send redirects.
1665 It is very good, but in some (rare!) circumstances
1666 (SIT, PtP, NBMA NOARP links) it is handy to allow
1667 some exceptions. --ANK
1668 */
1669 err = -EINVAL;
38308473 1670 if (!(gwa_type & IPV6_ADDR_UNICAST))
1da177e4
LT		 1671 goto out;
1672
5578689a 1673 grt = rt6_lookup(net, gw_addr, NULL, cfg->fc_ifindex, 1);
1da177e4
LT		 1674
1675 err = -EHOSTUNREACH;
38308473 1676 if (!grt)
1da177e4
LT		 1677 goto out;
1678 if (dev) {
d1918542 1679 if (dev != grt->dst.dev) {
94e187c0 1680 ip6_rt_put(grt);
1da177e4
LT		 1681 goto out;
1682 }
1683 } else {
d1918542 1684 dev = grt->dst.dev;
1da177e4
LT		 1685 idev = grt->rt6i_idev;
1686 dev_hold(dev);
1687 in6_dev_hold(grt->rt6i_idev);
1688 }
38308473 1689 if (!(grt->rt6i_flags & RTF_GATEWAY))
1da177e4 1690 err = 0;
94e187c0 1691 ip6_rt_put(grt);
1da177e4
LT		 1692
1693 if (err)
1694 goto out;
1695 }
1696 err = -EINVAL;
38308473 1697 if (!dev || (dev->flags & IFF_LOOPBACK))
1da177e4
LT		 1698 goto out;
1699 }
1700
1701 err = -ENODEV;
38308473 1702 if (!dev)
1da177e4
LT		 1703 goto out;
1704
c3968a85
DW		 1705 if (!ipv6_addr_any(&cfg->fc_prefsrc)) {
1706 if (!ipv6_chk_addr(net, &cfg->fc_prefsrc, dev, 0)) {
1707 err = -EINVAL;
1708 goto out;
1709 }
4e3fd7a0 1710 rt->rt6i_prefsrc.addr = cfg->fc_prefsrc;
c3968a85
DW		 1711 rt->rt6i_prefsrc.plen = 128;
1712 } else
1713 rt->rt6i_prefsrc.plen = 0;
1714
86872cb5 1715 rt->rt6i_flags = cfg->fc_flags;
1da177e4
LT		 1716
1717install_route:
d8d1f30b 1718 rt->dst.dev = dev;
1da177e4 1719 rt->rt6i_idev = idev;
c71099ac 1720 rt->rt6i_table = table;
63152fc0 1721
c346dca1 1722 cfg->fc_nlinfo.nl_net = dev_net(dev);
63152fc0 1723
e715b6d3
FW		 1724 err = ip6_convert_metrics(&mxc, cfg);
1725 if (err)
1726 goto out;
1da177e4 1727
e715b6d3
FW		 1728 err = __ip6_ins_rt(rt, &cfg->fc_nlinfo, &mxc);
1729
1730 kfree(mxc.mx);
1731 return err;
1da177e4
LT		 1732out:
1733 if (dev)
1734 dev_put(dev);
1735 if (idev)
1736 in6_dev_put(idev);
1737 if (rt)
d8d1f30b 1738 dst_free(&rt->dst);
1da177e4
LT		 1739 return err;
1740}
1741
86872cb5 1742static int __ip6_del_rt(struct rt6_info *rt, struct nl_info *info)
1da177e4
LT		 1743{
1744 int err;
c71099ac 1745 struct fib6_table *table;
d1918542 1746 struct net *net = dev_net(rt->dst.dev);
1da177e4 1747
6825a26c
G		 1748 if (rt == net->ipv6.ip6_null_entry) {
1749 err = -ENOENT;
1750 goto out;
1751 }
6c813a72 1752
c71099ac
TG		 1753 table = rt->rt6i_table;
1754 write_lock_bh(&table->tb6_lock);
86872cb5 1755 err = fib6_del(rt, info);
c71099ac 1756 write_unlock_bh(&table->tb6_lock);
1da177e4 1757
6825a26c 1758out:
94e187c0 1759 ip6_rt_put(rt);
1da177e4
LT		 1760 return err;
1761}
1762
e0a1ad73
TG		 1763int ip6_del_rt(struct rt6_info *rt)
1764{
4d1169c1 1765 struct nl_info info = {
d1918542 1766 .nl_net = dev_net(rt->dst.dev),
4d1169c1 1767 };
528c4ceb 1768 return __ip6_del_rt(rt, &info);
e0a1ad73
TG		 1769}
1770
86872cb5 1771static int ip6_route_del(struct fib6_config *cfg)
1da177e4 1772{
c71099ac 1773 struct fib6_table *table;
1da177e4
LT		 1774 struct fib6_node *fn;
1775 struct rt6_info *rt;
1776 int err = -ESRCH;
1777
5578689a 1778 table = fib6_get_table(cfg->fc_nlinfo.nl_net, cfg->fc_table);
38308473 1779 if (!table)
c71099ac
TG		 1780 return err;
1781
1782 read_lock_bh(&table->tb6_lock);
1da177e4 1783
c71099ac 1784 fn = fib6_locate(&table->tb6_root,
86872cb5
TG		 1785 &cfg->fc_dst, cfg->fc_dst_len,
1786 &cfg->fc_src, cfg->fc_src_len);
1ab1457c 1787
1da177e4 1788 if (fn) {
d8d1f30b 1789 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
86872cb5 1790 if (cfg->fc_ifindex &&
d1918542
DM		 1791 (!rt->dst.dev ||
1792 rt->dst.dev->ifindex != cfg->fc_ifindex))
1da177e4 1793 continue;
86872cb5
TG		 1794 if (cfg->fc_flags & RTF_GATEWAY &&
1795 !ipv6_addr_equal(&cfg->fc_gateway, &rt->rt6i_gateway))
1da177e4 1796 continue;
86872cb5 1797 if (cfg->fc_metric && cfg->fc_metric != rt->rt6i_metric)
1da177e4 1798 continue;
d8d1f30b 1799 dst_hold(&rt->dst);
c71099ac 1800 read_unlock_bh(&table->tb6_lock);
1da177e4 1801
86872cb5 1802 return __ip6_del_rt(rt, &cfg->fc_nlinfo);
1da177e4
LT		 1803 }
1804 }
c71099ac 1805 read_unlock_bh(&table->tb6_lock);
1da177e4
LT		 1806
1807 return err;
1808}
1809
6700c270 1810static void rt6_do_redirect(struct dst_entry *dst, struct sock *sk, struct sk_buff *skb)
a6279458 1811{
e8599ff4 1812 struct net *net = dev_net(skb->dev);
a6279458 1813 struct netevent_redirect netevent;
e8599ff4 1814 struct rt6_info *rt, *nrt = NULL;
e8599ff4
DM		 1815 struct ndisc_options ndopts;
1816 struct inet6_dev *in6_dev;
1817 struct neighbour *neigh;
71bcdba0 1818 struct rd_msg *msg;
6e157b6a
DM		 1819 int optlen, on_link;
1820 u8 *lladdr;
e8599ff4 1821
29a3cad5 1822 optlen = skb_tail_pointer(skb) - skb_transport_header(skb);
71bcdba0 1823 optlen -= sizeof(*msg);
e8599ff4
DM		 1824
1825 if (optlen < 0) {
6e157b6a 1826 net_dbg_ratelimited("rt6_do_redirect: packet too short\n");
e8599ff4
DM		 1827 return;
1828 }
1829
71bcdba0 1830 msg = (struct rd_msg *)icmp6_hdr(skb);
e8599ff4 1831
71bcdba0 1832 if (ipv6_addr_is_multicast(&msg->dest)) {
6e157b6a 1833 net_dbg_ratelimited("rt6_do_redirect: destination address is multicast\n");
e8599ff4
DM		 1834 return;
1835 }
1836
6e157b6a 1837 on_link = 0;
71bcdba0 1838 if (ipv6_addr_equal(&msg->dest, &msg->target)) {
e8599ff4 1839 on_link = 1;
71bcdba0 1840 } else if (ipv6_addr_type(&msg->target) !=
e8599ff4 1841 (IPV6_ADDR_UNICAST|IPV6_ADDR_LINKLOCAL)) {
6e157b6a 1842 net_dbg_ratelimited("rt6_do_redirect: target address is not link-local unicast\n");
e8599ff4
DM		 1843 return;
1844 }
1845
1846 in6_dev = __in6_dev_get(skb->dev);
1847 if (!in6_dev)
1848 return;
1849 if (in6_dev->cnf.forwarding || !in6_dev->cnf.accept_redirects)
1850 return;
1851
1852 /* RFC2461 8.1:
1853 * The IP source address of the Redirect MUST be the same as the current
1854 * first-hop router for the specified ICMP Destination Address.
1855 */
1856
71bcdba0 1857 if (!ndisc_parse_options(msg->opt, optlen, &ndopts)) {
e8599ff4
DM		 1858 net_dbg_ratelimited("rt6_redirect: invalid ND options\n");
1859 return;
1860 }
6e157b6a
DM		 1861
1862 lladdr = NULL;
e8599ff4
DM		 1863 if (ndopts.nd_opts_tgt_lladdr) {
1864 lladdr = ndisc_opt_addr_data(ndopts.nd_opts_tgt_lladdr,
1865 skb->dev);
1866 if (!lladdr) {
1867 net_dbg_ratelimited("rt6_redirect: invalid link-layer address length\n");
1868 return;
1869 }
1870 }
1871
6e157b6a
DM		 1872 rt = (struct rt6_info *) dst;
1873 if (rt == net->ipv6.ip6_null_entry) {
1874 net_dbg_ratelimited("rt6_redirect: source isn't a valid nexthop for redirect target\n");
e8599ff4 1875 return;
6e157b6a 1876 }
e8599ff4 1877
6e157b6a
DM		 1878 /* Redirect received -> path was valid.
1879 * Look, redirects are sent only in response to data packets,
1880 * so that this nexthop apparently is reachable. --ANK
1881 */
1882 dst_confirm(&rt->dst);
a6279458 1883
71bcdba0 1884 neigh = __neigh_lookup(&nd_tbl, &msg->target, skb->dev, 1);
6e157b6a
DM		 1885 if (!neigh)
1886 return;
a6279458 1887
1da177e4
LT		 1888 /*
1889 * We have finally decided to accept it.
1890 */
1891
1ab1457c 1892 neigh_update(neigh, lladdr, NUD_STALE,
1da177e4
LT		 1893 NEIGH_UPDATE_F_WEAK_OVERRIDE|
1894 NEIGH_UPDATE_F_OVERRIDE|
1895 (on_link ? 0 : (NEIGH_UPDATE_F_OVERRIDE_ISROUTER|
1896 NEIGH_UPDATE_F_ISROUTER))
1897 );
1898
71bcdba0 1899 nrt = ip6_rt_copy(rt, &msg->dest);
38308473 1900 if (!nrt)
1da177e4
LT		 1901 goto out;
1902
1903 nrt->rt6i_flags = RTF_GATEWAY|RTF_UP|RTF_DYNAMIC|RTF_CACHE;
1904 if (on_link)
1905 nrt->rt6i_flags &= ~RTF_GATEWAY;
1906
4e3fd7a0 1907 nrt->rt6i_gateway = *(struct in6_addr *)neigh->primary_key;
1da177e4 1908
40e22e8f 1909 if (ip6_ins_rt(nrt))
1da177e4
LT		 1910 goto out;
1911
d8d1f30b
CG		 1912 netevent.old = &rt->dst;
1913 netevent.new = &nrt->dst;
71bcdba0 1914 netevent.daddr = &msg->dest;
60592833 1915 netevent.neigh = neigh;
8d71740c
TT		 1916 call_netevent_notifiers(NETEVENT_REDIRECT, &netevent);
1917
38308473 1918 if (rt->rt6i_flags & RTF_CACHE) {
6e157b6a 1919 rt = (struct rt6_info *) dst_clone(&rt->dst);
e0a1ad73 1920 ip6_del_rt(rt);
1da177e4
LT		 1921 }
1922
1923out:
e8599ff4 1924 neigh_release(neigh);
6e157b6a
DM		 1925}
1926
1da177e4
LT		 1927/*
1928 * Misc support functions
1929 */
1930
1716a961 1931static struct rt6_info *ip6_rt_copy(struct rt6_info *ort,
21efcfa0 1932 const struct in6_addr *dest)
1da177e4 1933{
d1918542 1934 struct net *net = dev_net(ort->dst.dev);
8b96d22d
DM		 1935 struct rt6_info *rt = ip6_dst_alloc(net, ort->dst.dev, 0,
1936 ort->rt6i_table);
1da177e4
LT		 1937
1938 if (rt) {
d8d1f30b
CG		 1939 rt->dst.input = ort->dst.input;
1940 rt->dst.output = ort->dst.output;
8e2ec639 1941 rt->dst.flags |= DST_HOST;
d8d1f30b 1942
4e3fd7a0 1943 rt->rt6i_dst.addr = *dest;
8e2ec639 1944 rt->rt6i_dst.plen = 128;
defb3519 1945 dst_copy_metrics(&rt->dst, &ort->dst);
d8d1f30b 1946 rt->dst.error = ort->dst.error;
1da177e4
LT		 1947 rt->rt6i_idev = ort->rt6i_idev;
1948 if (rt->rt6i_idev)
1949 in6_dev_hold(rt->rt6i_idev);
d8d1f30b 1950 rt->dst.lastuse = jiffies;
1da177e4 1951
550bab42
JA		 1952 if (ort->rt6i_flags & RTF_GATEWAY)
1953 rt->rt6i_gateway = ort->rt6i_gateway;
1954 else
1955 rt->rt6i_gateway = *dest;
1716a961 1956 rt->rt6i_flags = ort->rt6i_flags;
24f5b855 1957 rt6_set_from(rt, ort);
1da177e4
LT		 1958 rt->rt6i_metric = 0;
1959
1da177e4
LT		 1960#ifdef CONFIG_IPV6_SUBTREES
1961 memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key));
1962#endif
0f6c6392 1963 memcpy(&rt->rt6i_prefsrc, &ort->rt6i_prefsrc, sizeof(struct rt6key));
c71099ac 1964 rt->rt6i_table = ort->rt6i_table;
1da177e4
LT		 1965 }
1966 return rt;
1967}
1968
70ceb4f5 1969#ifdef CONFIG_IPV6_ROUTE_INFO
efa2cea0 1970static struct rt6_info *rt6_get_route_info(struct net *net,
b71d1d42
ED		 1971 const struct in6_addr *prefix, int prefixlen,
1972 const struct in6_addr *gwaddr, int ifindex)
70ceb4f5
YH		 1973{
1974 struct fib6_node *fn;
1975 struct rt6_info *rt = NULL;
c71099ac
TG		 1976 struct fib6_table *table;
1977
efa2cea0 1978 table = fib6_get_table(net, RT6_TABLE_INFO);
38308473 1979 if (!table)
c71099ac 1980 return NULL;
70ceb4f5 1981
5744dd9b 1982 read_lock_bh(&table->tb6_lock);
67ba4152 1983 fn = fib6_locate(&table->tb6_root, prefix, prefixlen, NULL, 0);
70ceb4f5
YH		 1984 if (!fn)
1985 goto out;
1986
d8d1f30b 1987 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
d1918542 1988 if (rt->dst.dev->ifindex != ifindex)
70ceb4f5
YH		 1989 continue;
1990 if ((rt->rt6i_flags & (RTF_ROUTEINFO|RTF_GATEWAY)) != (RTF_ROUTEINFO|RTF_GATEWAY))
1991 continue;
1992 if (!ipv6_addr_equal(&rt->rt6i_gateway, gwaddr))
1993 continue;
d8d1f30b 1994 dst_hold(&rt->dst);
70ceb4f5
YH		 1995 break;
1996 }
1997out:
5744dd9b 1998 read_unlock_bh(&table->tb6_lock);
70ceb4f5
YH		 1999 return rt;
2000}
2001
efa2cea0 2002static struct rt6_info *rt6_add_route_info(struct net *net,
b71d1d42
ED		 2003 const struct in6_addr *prefix, int prefixlen,
2004 const struct in6_addr *gwaddr, int ifindex,
95c96174 2005 unsigned int pref)
70ceb4f5 2006{
86872cb5
TG		 2007 struct fib6_config cfg = {
2008 .fc_table = RT6_TABLE_INFO,
238fc7ea 2009 .fc_metric = IP6_RT_PRIO_USER,
86872cb5
TG		 2010 .fc_ifindex = ifindex,
2011 .fc_dst_len = prefixlen,
2012 .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_ROUTEINFO |
2013 RTF_UP | RTF_PREF(pref),
15e47304 2014 .fc_nlinfo.portid = 0,
efa2cea0
DL		 2015 .fc_nlinfo.nlh = NULL,
2016 .fc_nlinfo.nl_net = net,
86872cb5
TG		 2017 };
2018
4e3fd7a0
AD		 2019 cfg.fc_dst = *prefix;
2020 cfg.fc_gateway = *gwaddr;
70ceb4f5 2021
e317da96
YH		 2022 /* We should treat it as a default route if prefix length is 0. */
2023 if (!prefixlen)
86872cb5 2024 cfg.fc_flags |= RTF_DEFAULT;
70ceb4f5 2025
86872cb5 2026 ip6_route_add(&cfg);
70ceb4f5 2027
efa2cea0 2028 return rt6_get_route_info(net, prefix, prefixlen, gwaddr, ifindex);
70ceb4f5
YH		 2029}
2030#endif
2031
b71d1d42 2032struct rt6_info *rt6_get_dflt_router(const struct in6_addr *addr, struct net_device *dev)
1ab1457c 2033{
1da177e4 2034 struct rt6_info *rt;
c71099ac 2035 struct fib6_table *table;
1da177e4 2036
c346dca1 2037 table = fib6_get_table(dev_net(dev), RT6_TABLE_DFLT);
38308473 2038 if (!table)
c71099ac 2039 return NULL;
1da177e4 2040
5744dd9b 2041 read_lock_bh(&table->tb6_lock);
67ba4152 2042 for (rt = table->tb6_root.leaf; rt; rt = rt->dst.rt6_next) {
d1918542 2043 if (dev == rt->dst.dev &&
045927ff 2044 ((rt->rt6i_flags & (RTF_ADDRCONF | RTF_DEFAULT)) == (RTF_ADDRCONF | RTF_DEFAULT)) &&
1da177e4
LT		 2045 ipv6_addr_equal(&rt->rt6i_gateway, addr))
2046 break;
2047 }
2048 if (rt)
d8d1f30b 2049 dst_hold(&rt->dst);
5744dd9b 2050 read_unlock_bh(&table->tb6_lock);
1da177e4
LT		 2051 return rt;
2052}
2053
b71d1d42 2054struct rt6_info *rt6_add_dflt_router(const struct in6_addr *gwaddr,
ebacaaa0
YH		 2055 struct net_device *dev,
2056 unsigned int pref)
1da177e4 2057{
86872cb5
TG		 2058 struct fib6_config cfg = {
2059 .fc_table = RT6_TABLE_DFLT,
238fc7ea 2060 .fc_metric = IP6_RT_PRIO_USER,
86872cb5
TG		 2061 .fc_ifindex = dev->ifindex,
2062 .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_DEFAULT |
2063 RTF_UP | RTF_EXPIRES | RTF_PREF(pref),
15e47304 2064 .fc_nlinfo.portid = 0,
5578689a 2065 .fc_nlinfo.nlh = NULL,
c346dca1 2066 .fc_nlinfo.nl_net = dev_net(dev),
86872cb5 2067 };
1da177e4 2068
4e3fd7a0 2069 cfg.fc_gateway = *gwaddr;
1da177e4 2070
86872cb5 2071 ip6_route_add(&cfg);
1da177e4 2072
1da177e4
LT		 2073 return rt6_get_dflt_router(gwaddr, dev);
2074}
2075
7b4da532 2076void rt6_purge_dflt_routers(struct net *net)
1da177e4
LT		 2077{
2078 struct rt6_info *rt;
c71099ac
TG		 2079 struct fib6_table *table;
2080
2081 /* NOTE: Keep consistent with rt6_get_dflt_router */
7b4da532 2082 table = fib6_get_table(net, RT6_TABLE_DFLT);
38308473 2083 if (!table)
c71099ac 2084 return;
1da177e4
LT		 2085
2086restart:
c71099ac 2087 read_lock_bh(&table->tb6_lock);
d8d1f30b 2088 for (rt = table->tb6_root.leaf; rt; rt = rt->dst.rt6_next) {
3e8b0ac3
LC		 2089 if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ADDRCONF) &&
2090 (!rt->rt6i_idev || rt->rt6i_idev->cnf.accept_ra != 2)) {
d8d1f30b 2091 dst_hold(&rt->dst);
c71099ac 2092 read_unlock_bh(&table->tb6_lock);
e0a1ad73 2093 ip6_del_rt(rt);
1da177e4
LT		 2094 goto restart;
2095 }
2096 }
c71099ac 2097 read_unlock_bh(&table->tb6_lock);
1da177e4
LT		 2098}
2099
5578689a
DL		 2100static void rtmsg_to_fib6_config(struct net *net,
2101 struct in6_rtmsg *rtmsg,
86872cb5
TG		 2102 struct fib6_config *cfg)
2103{
2104 memset(cfg, 0, sizeof(*cfg));
2105
2106 cfg->fc_table = RT6_TABLE_MAIN;
2107 cfg->fc_ifindex = rtmsg->rtmsg_ifindex;
2108 cfg->fc_metric = rtmsg->rtmsg_metric;
2109 cfg->fc_expires = rtmsg->rtmsg_info;
2110 cfg->fc_dst_len = rtmsg->rtmsg_dst_len;
2111 cfg->fc_src_len = rtmsg->rtmsg_src_len;
2112 cfg->fc_flags = rtmsg->rtmsg_flags;
2113
5578689a 2114 cfg->fc_nlinfo.nl_net = net;
f1243c2d 2115
4e3fd7a0
AD		 2116 cfg->fc_dst = rtmsg->rtmsg_dst;
2117 cfg->fc_src = rtmsg->rtmsg_src;
2118 cfg->fc_gateway = rtmsg->rtmsg_gateway;
86872cb5
TG		 2119}
2120
5578689a 2121int ipv6_route_ioctl(struct net *net, unsigned int cmd, void __user *arg)
1da177e4 2122{
86872cb5 2123 struct fib6_config cfg;
1da177e4
LT		 2124 struct in6_rtmsg rtmsg;
2125 int err;
2126
67ba4152 2127 switch (cmd) {
1da177e4
LT		 2128 case SIOCADDRT: /* Add a route */
2129 case SIOCDELRT: /* Delete a route */
af31f412 2130 if (!ns_capable(net->user_ns, CAP_NET_ADMIN))
1da177e4
LT		 2131 return -EPERM;
2132 err = copy_from_user(&rtmsg, arg,
2133 sizeof(struct in6_rtmsg));
2134 if (err)
2135 return -EFAULT;
86872cb5 2136
5578689a 2137 rtmsg_to_fib6_config(net, &rtmsg, &cfg);
86872cb5 2138
1da177e4
LT		 2139 rtnl_lock();
2140 switch (cmd) {
2141 case SIOCADDRT:
86872cb5 2142 err = ip6_route_add(&cfg);
1da177e4
LT		 2143 break;
2144 case SIOCDELRT:
86872cb5 2145 err = ip6_route_del(&cfg);
1da177e4
LT		 2146 break;
2147 default:
2148 err = -EINVAL;
2149 }
2150 rtnl_unlock();
2151
2152 return err;
3ff50b79 2153 }
1da177e4
LT		 2154
2155 return -EINVAL;
2156}
2157
2158/*
2159 * Drop the packet on the floor
2160 */
2161
d5fdd6ba 2162static int ip6_pkt_drop(struct sk_buff *skb, u8 code, int ipstats_mib_noroutes)
1da177e4 2163{
612f09e8 2164 int type;
adf30907 2165 struct dst_entry *dst = skb_dst(skb);
612f09e8
YH		 2166 switch (ipstats_mib_noroutes) {
2167 case IPSTATS_MIB_INNOROUTES:
0660e03f 2168 type = ipv6_addr_type(&ipv6_hdr(skb)->daddr);
45bb0060 2169 if (type == IPV6_ADDR_ANY) {
3bd653c8
DL		 2170 IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst),
2171 IPSTATS_MIB_INADDRERRORS);
612f09e8
YH		 2172 break;
2173 }
2174 /* FALLTHROUGH */
2175 case IPSTATS_MIB_OUTNOROUTES:
3bd653c8
DL		 2176 IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst),
2177 ipstats_mib_noroutes);
612f09e8
YH		 2178 break;
2179 }
3ffe533c 2180 icmpv6_send(skb, ICMPV6_DEST_UNREACH, code, 0);
1da177e4
LT		 2181 kfree_skb(skb);
2182 return 0;
2183}
2184
9ce8ade0
TG		 2185static int ip6_pkt_discard(struct sk_buff *skb)
2186{
612f09e8 2187 return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_INNOROUTES);
9ce8ade0
TG		 2188}
2189
aad88724 2190static int ip6_pkt_discard_out(struct sock *sk, struct sk_buff *skb)
1da177e4 2191{
adf30907 2192 skb->dev = skb_dst(skb)->dev;
612f09e8 2193 return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_OUTNOROUTES);
1da177e4
LT		 2194}
2195
9ce8ade0
TG		 2196static int ip6_pkt_prohibit(struct sk_buff *skb)
2197{
612f09e8 2198 return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_INNOROUTES);
9ce8ade0
TG		 2199}
2200
aad88724 2201static int ip6_pkt_prohibit_out(struct sock *sk, struct sk_buff *skb)
9ce8ade0 2202{
adf30907 2203 skb->dev = skb_dst(skb)->dev;
612f09e8 2204 return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_OUTNOROUTES);
9ce8ade0
TG		 2205}
2206
1da177e4
LT		 2207/*
2208 * Allocate a dst for local (unicast / anycast) address.
2209 */
2210
2211struct rt6_info *addrconf_dst_alloc(struct inet6_dev *idev,
2212 const struct in6_addr *addr,
8f031519 2213 bool anycast)
1da177e4 2214{
c346dca1 2215 struct net *net = dev_net(idev->dev);
a3300ef4
HFS		 2216 struct rt6_info *rt = ip6_dst_alloc(net, net->loopback_dev,
2217 DST_NOCOUNT, NULL);
2218 if (!rt)
1da177e4
LT		 2219 return ERR_PTR(-ENOMEM);
2220
1da177e4
LT		 2221 in6_dev_hold(idev);
2222
11d53b49 2223 rt->dst.flags |= DST_HOST;
d8d1f30b
CG		 2224 rt->dst.input = ip6_input;
2225 rt->dst.output = ip6_output;
1da177e4 2226 rt->rt6i_idev = idev;
1da177e4
LT		 2227
2228 rt->rt6i_flags = RTF_UP | RTF_NONEXTHOP;
58c4fb86
YH		 2229 if (anycast)
2230 rt->rt6i_flags |= RTF_ANYCAST;
2231 else
1da177e4 2232 rt->rt6i_flags |= RTF_LOCAL;
1da177e4 2233
550bab42 2234 rt->rt6i_gateway = *addr;
4e3fd7a0 2235 rt->rt6i_dst.addr = *addr;
1da177e4 2236 rt->rt6i_dst.plen = 128;
5578689a 2237 rt->rt6i_table = fib6_get_table(net, RT6_TABLE_LOCAL);
1da177e4 2238
d8d1f30b 2239 atomic_set(&rt->dst.__refcnt, 1);
1da177e4
LT		 2240
2241 return rt;
2242}
2243
c3968a85
DW		 2244int ip6_route_get_saddr(struct net *net,
2245 struct rt6_info *rt,
b71d1d42 2246 const struct in6_addr *daddr,
c3968a85
DW		 2247 unsigned int prefs,
2248 struct in6_addr *saddr)
2249{
67ba4152 2250 struct inet6_dev *idev = ip6_dst_idev((struct dst_entry *)rt);
c3968a85
DW		 2251 int err = 0;
2252 if (rt->rt6i_prefsrc.plen)
4e3fd7a0 2253 *saddr = rt->rt6i_prefsrc.addr;
c3968a85
DW		 2254 else
2255 err = ipv6_dev_get_saddr(net, idev ? idev->dev : NULL,
2256 daddr, prefs, saddr);
2257 return err;
2258}
2259
2260/* remove deleted ip from prefsrc entries */
2261struct arg_dev_net_ip {
2262 struct net_device *dev;
2263 struct net *net;
2264 struct in6_addr *addr;
2265};
2266
2267static int fib6_remove_prefsrc(struct rt6_info *rt, void *arg)
2268{
2269 struct net_device *dev = ((struct arg_dev_net_ip *)arg)->dev;
2270 struct net *net = ((struct arg_dev_net_ip *)arg)->net;
2271 struct in6_addr *addr = ((struct arg_dev_net_ip *)arg)->addr;
2272
d1918542 2273 if (((void *)rt->dst.dev == dev || !dev) &&
c3968a85
DW		 2274 rt != net->ipv6.ip6_null_entry &&
2275 ipv6_addr_equal(addr, &rt->rt6i_prefsrc.addr)) {
2276 /* remove prefsrc entry */
2277 rt->rt6i_prefsrc.plen = 0;
2278 }
2279 return 0;
2280}
2281
2282void rt6_remove_prefsrc(struct inet6_ifaddr *ifp)
2283{
2284 struct net *net = dev_net(ifp->idev->dev);
2285 struct arg_dev_net_ip adni = {
2286 .dev = ifp->idev->dev,
2287 .net = net,
2288 .addr = &ifp->addr,
2289 };
0c3584d5 2290 fib6_clean_all(net, fib6_remove_prefsrc, &adni);
c3968a85
DW		 2291}
2292
be7a010d
DJ		 2293#define RTF_RA_ROUTER (RTF_ADDRCONF | RTF_DEFAULT | RTF_GATEWAY)
2294#define RTF_CACHE_GATEWAY (RTF_GATEWAY | RTF_CACHE)
2295
2296/* Remove routers and update dst entries when gateway turn into host. */
2297static int fib6_clean_tohost(struct rt6_info *rt, void *arg)
2298{
2299 struct in6_addr *gateway = (struct in6_addr *)arg;
2300
2301 if ((((rt->rt6i_flags & RTF_RA_ROUTER) == RTF_RA_ROUTER) ||
2302 ((rt->rt6i_flags & RTF_CACHE_GATEWAY) == RTF_CACHE_GATEWAY)) &&
2303 ipv6_addr_equal(gateway, &rt->rt6i_gateway)) {
2304 return -1;
2305 }
2306 return 0;
2307}
2308
2309void rt6_clean_tohost(struct net *net, struct in6_addr *gateway)
2310{
2311 fib6_clean_all(net, fib6_clean_tohost, gateway);
2312}
2313
8ed67789
DL		 2314struct arg_dev_net {
2315 struct net_device *dev;
2316 struct net *net;
2317};
2318
1da177e4
LT		 2319static int fib6_ifdown(struct rt6_info *rt, void *arg)
2320{
bc3ef660 2321 const struct arg_dev_net *adn = arg;
2322 const struct net_device *dev = adn->dev;
8ed67789 2323
d1918542 2324 if ((rt->dst.dev == dev || !dev) &&
c159d30c 2325 rt != adn->net->ipv6.ip6_null_entry)
1da177e4 2326 return -1;
c159d30c 2327
1da177e4
LT		 2328 return 0;
2329}
2330
f3db4851 2331void rt6_ifdown(struct net *net, struct net_device *dev)
1da177e4 2332{
8ed67789
DL		 2333 struct arg_dev_net adn = {
2334 .dev = dev,
2335 .net = net,
2336 };
2337
0c3584d5 2338 fib6_clean_all(net, fib6_ifdown, &adn);
1e493d19 2339 icmp6_clean_all(fib6_ifdown, &adn);
1da177e4
LT		 2340}
2341
95c96174 2342struct rt6_mtu_change_arg {
1da177e4 2343 struct net_device *dev;
95c96174 2344 unsigned int mtu;
1da177e4
LT		 2345};
2346
2347static int rt6_mtu_change_route(struct rt6_info *rt, void *p_arg)
2348{
2349 struct rt6_mtu_change_arg *arg = (struct rt6_mtu_change_arg *) p_arg;
2350 struct inet6_dev *idev;
2351
2352 /* In IPv6 pmtu discovery is not optional,
2353 so that RTAX_MTU lock cannot disable it.
2354 We still use this lock to block changes
2355 caused by addrconf/ndisc.
2356 */
2357
2358 idev = __in6_dev_get(arg->dev);
38308473 2359 if (!idev)
1da177e4
LT		 2360 return 0;
2361
2362 /* For administrative MTU increase, there is no way to discover
2363 IPv6 PMTU increase, so PMTU increase should be updated here.
2364 Since RFC 1981 doesn't include administrative MTU increase
2365 update PMTU increase is a MUST. (i.e. jumbo frame)
2366 */
2367 /*
2368 If new MTU is less than route PMTU, this new MTU will be the
2369 lowest MTU in the path, update the route PMTU to reflect PMTU
2370 decreases; if new MTU is greater than route PMTU, and the
2371 old MTU is the lowest MTU in the path, update the route PMTU
2372 to reflect the increase. In this case if the other nodes' MTU
2373 also have the lowest MTU, TOO BIG MESSAGE will be lead to
2374 PMTU discouvery.
2375 */
d1918542 2376 if (rt->dst.dev == arg->dev &&
d8d1f30b
CG		 2377 !dst_metric_locked(&rt->dst, RTAX_MTU) &&
2378 (dst_mtu(&rt->dst) >= arg->mtu ||
2379 (dst_mtu(&rt->dst) < arg->mtu &&
2380 dst_mtu(&rt->dst) == idev->cnf.mtu6))) {
defb3519 2381 dst_metric_set(&rt->dst, RTAX_MTU, arg->mtu);
566cfd8f 2382 }
1da177e4
LT		 2383 return 0;
2384}
2385
95c96174 2386void rt6_mtu_change(struct net_device *dev, unsigned int mtu)
1da177e4 2387{
c71099ac
TG		 2388 struct rt6_mtu_change_arg arg = {
2389 .dev = dev,
2390 .mtu = mtu,
2391 };
1da177e4 2392
0c3584d5 2393 fib6_clean_all(dev_net(dev), rt6_mtu_change_route, &arg);
1da177e4
LT		 2394}
2395
ef7c79ed 2396static const struct nla_policy rtm_ipv6_policy[RTA_MAX+1] = {
5176f91e 2397 [RTA_GATEWAY] = { .len = sizeof(struct in6_addr) },
86872cb5 2398 [RTA_OIF] = { .type = NLA_U32 },
ab364a6f 2399 [RTA_IIF] = { .type = NLA_U32 },
86872cb5
TG		 2400 [RTA_PRIORITY] = { .type = NLA_U32 },
2401 [RTA_METRICS] = { .type = NLA_NESTED },
51ebd318 2402 [RTA_MULTIPATH] = { .len = sizeof(struct rtnexthop) },
86872cb5
TG		 2403};
2404
2405static int rtm_to_fib6_config(struct sk_buff *skb, struct nlmsghdr *nlh,
2406 struct fib6_config *cfg)
1da177e4 2407{
86872cb5
TG		 2408 struct rtmsg *rtm;
2409 struct nlattr *tb[RTA_MAX+1];
2410 int err;
1da177e4 2411
86872cb5
TG		 2412 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy);
2413 if (err < 0)
2414 goto errout;
1da177e4 2415
86872cb5
TG		 2416 err = -EINVAL;
2417 rtm = nlmsg_data(nlh);
2418 memset(cfg, 0, sizeof(*cfg));
2419
2420 cfg->fc_table = rtm->rtm_table;
2421 cfg->fc_dst_len = rtm->rtm_dst_len;
2422 cfg->fc_src_len = rtm->rtm_src_len;
2423 cfg->fc_flags = RTF_UP;
2424 cfg->fc_protocol = rtm->rtm_protocol;
ef2c7d7b 2425 cfg->fc_type = rtm->rtm_type;
86872cb5 2426
ef2c7d7b
ND		 2427 if (rtm->rtm_type == RTN_UNREACHABLE ||
2428 rtm->rtm_type == RTN_BLACKHOLE ||
b4949ab2
ND		 2429 rtm->rtm_type == RTN_PROHIBIT ||
2430 rtm->rtm_type == RTN_THROW)
86872cb5
TG		 2431 cfg->fc_flags |= RTF_REJECT;
2432
ab79ad14
MŻ		 2433 if (rtm->rtm_type == RTN_LOCAL)
2434 cfg->fc_flags |= RTF_LOCAL;
2435
15e47304 2436 cfg->fc_nlinfo.portid = NETLINK_CB(skb).portid;
86872cb5 2437 cfg->fc_nlinfo.nlh = nlh;
3b1e0a65 2438 cfg->fc_nlinfo.nl_net = sock_net(skb->sk);
86872cb5
TG		 2439
2440 if (tb[RTA_GATEWAY]) {
2441 nla_memcpy(&cfg->fc_gateway, tb[RTA_GATEWAY], 16);
2442 cfg->fc_flags |= RTF_GATEWAY;
1da177e4 2443 }
86872cb5
TG		 2444
2445 if (tb[RTA_DST]) {
2446 int plen = (rtm->rtm_dst_len + 7) >> 3;
2447
2448 if (nla_len(tb[RTA_DST]) < plen)
2449 goto errout;
2450
2451 nla_memcpy(&cfg->fc_dst, tb[RTA_DST], plen);
1da177e4 2452 }
86872cb5
TG		 2453
2454 if (tb[RTA_SRC]) {
2455 int plen = (rtm->rtm_src_len + 7) >> 3;
2456
2457 if (nla_len(tb[RTA_SRC]) < plen)
2458 goto errout;
2459
2460 nla_memcpy(&cfg->fc_src, tb[RTA_SRC], plen);
1da177e4 2461 }
86872cb5 2462
c3968a85
DW		 2463 if (tb[RTA_PREFSRC])
2464 nla_memcpy(&cfg->fc_prefsrc, tb[RTA_PREFSRC], 16);
2465
86872cb5
TG		 2466 if (tb[RTA_OIF])
2467 cfg->fc_ifindex = nla_get_u32(tb[RTA_OIF]);
2468
2469 if (tb[RTA_PRIORITY])
2470 cfg->fc_metric = nla_get_u32(tb[RTA_PRIORITY]);
2471
2472 if (tb[RTA_METRICS]) {
2473 cfg->fc_mx = nla_data(tb[RTA_METRICS]);
2474 cfg->fc_mx_len = nla_len(tb[RTA_METRICS]);
1da177e4 2475 }
86872cb5
TG		 2476
2477 if (tb[RTA_TABLE])
2478 cfg->fc_table = nla_get_u32(tb[RTA_TABLE]);
2479
51ebd318
ND		 2480 if (tb[RTA_MULTIPATH]) {
2481 cfg->fc_mp = nla_data(tb[RTA_MULTIPATH]);
2482 cfg->fc_mp_len = nla_len(tb[RTA_MULTIPATH]);
2483 }
2484
86872cb5
TG		 2485 err = 0;
2486errout:
2487 return err;
1da177e4
LT		 2488}
2489
51ebd318
ND		 2490static int ip6_route_multipath(struct fib6_config *cfg, int add)
2491{
2492 struct fib6_config r_cfg;
2493 struct rtnexthop *rtnh;
2494 int remaining;
2495 int attrlen;
2496 int err = 0, last_err = 0;
2497
2498beginning:
2499 rtnh = (struct rtnexthop *)cfg->fc_mp;
2500 remaining = cfg->fc_mp_len;
2501
2502 /* Parse a Multipath Entry */
2503 while (rtnh_ok(rtnh, remaining)) {
2504 memcpy(&r_cfg, cfg, sizeof(*cfg));
2505 if (rtnh->rtnh_ifindex)
2506 r_cfg.fc_ifindex = rtnh->rtnh_ifindex;
2507
2508 attrlen = rtnh_attrlen(rtnh);
2509 if (attrlen > 0) {
2510 struct nlattr *nla, *attrs = rtnh_attrs(rtnh);
2511
2512 nla = nla_find(attrs, attrlen, RTA_GATEWAY);
2513 if (nla) {
2514 nla_memcpy(&r_cfg.fc_gateway, nla, 16);
2515 r_cfg.fc_flags |= RTF_GATEWAY;
2516 }
2517 }
2518 err = add ? ip6_route_add(&r_cfg) : ip6_route_del(&r_cfg);
2519 if (err) {
2520 last_err = err;
2521 /* If we are trying to remove a route, do not stop the
2522 * loop when ip6_route_del() fails (because next hop is
2523 * already gone), we should try to remove all next hops.
2524 */
2525 if (add) {
2526 /* If add fails, we should try to delete all
2527 * next hops that have been already added.
2528 */
2529 add = 0;
2530 goto beginning;
2531 }
2532 }
1a72418b
ND		 2533 /* Because each route is added like a single route we remove
2534 * this flag after the first nexthop (if there is a collision,
2535 * we have already fail to add the first nexthop:
2536 * fib6_add_rt2node() has reject it).
2537 */
2538 cfg->fc_nlinfo.nlh->nlmsg_flags &= ~NLM_F_EXCL;
51ebd318
ND		 2539 rtnh = rtnh_next(rtnh, &remaining);
2540 }
2541
2542 return last_err;
2543}
2544
67ba4152 2545static int inet6_rtm_delroute(struct sk_buff *skb, struct nlmsghdr *nlh)
1da177e4 2546{
86872cb5
TG		 2547 struct fib6_config cfg;
2548 int err;
1da177e4 2549
86872cb5
TG		 2550 err = rtm_to_fib6_config(skb, nlh, &cfg);
2551 if (err < 0)
2552 return err;
2553
51ebd318
ND		 2554 if (cfg.fc_mp)
2555 return ip6_route_multipath(&cfg, 0);
2556 else
2557 return ip6_route_del(&cfg);
1da177e4
LT		 2558}
2559
67ba4152 2560static int inet6_rtm_newroute(struct sk_buff *skb, struct nlmsghdr *nlh)
1da177e4 2561{
86872cb5
TG		 2562 struct fib6_config cfg;
2563 int err;
1da177e4 2564
86872cb5
TG		 2565 err = rtm_to_fib6_config(skb, nlh, &cfg);
2566 if (err < 0)
2567 return err;
2568
51ebd318
ND		 2569 if (cfg.fc_mp)
2570 return ip6_route_multipath(&cfg, 1);
2571 else
2572 return ip6_route_add(&cfg);
1da177e4
LT		 2573}
2574
339bf98f
TG		 2575static inline size_t rt6_nlmsg_size(void)
2576{
2577 return NLMSG_ALIGN(sizeof(struct rtmsg))
2578 + nla_total_size(16) /* RTA_SRC */
2579 + nla_total_size(16) /* RTA_DST */
2580 + nla_total_size(16) /* RTA_GATEWAY */
2581 + nla_total_size(16) /* RTA_PREFSRC */
2582 + nla_total_size(4) /* RTA_TABLE */
2583 + nla_total_size(4) /* RTA_IIF */
2584 + nla_total_size(4) /* RTA_OIF */
2585 + nla_total_size(4) /* RTA_PRIORITY */
6a2b9ce0 2586 + RTAX_MAX * nla_total_size(4) /* RTA_METRICS */
ea697639
DB		 2587 + nla_total_size(sizeof(struct rta_cacheinfo))
2588 + nla_total_size(TCP_CA_NAME_MAX); /* RTAX_CC_ALGO */
339bf98f
TG		 2589}
2590
191cd582
BH		 2591static int rt6_fill_node(struct net *net,
2592 struct sk_buff *skb, struct rt6_info *rt,
0d51aa80 2593 struct in6_addr *dst, struct in6_addr *src,
15e47304 2594 int iif, int type, u32 portid, u32 seq,
7bc570c8 2595 int prefix, int nowait, unsigned int flags)
1da177e4
LT		 2596{
2597 struct rtmsg *rtm;
2d7202bf 2598 struct nlmsghdr *nlh;
e3703b3d 2599 long expires;
9e762a4a 2600 u32 table;
1da177e4
LT		 2601
2602 if (prefix) { /* user wants prefix routes only */
2603 if (!(rt->rt6i_flags & RTF_PREFIX_RT)) {
2604 /* success since this is not a prefix route */
2605 return 1;
2606 }
2607 }
2608
15e47304 2609 nlh = nlmsg_put(skb, portid, seq, type, sizeof(*rtm), flags);
38308473 2610 if (!nlh)
26932566 2611 return -EMSGSIZE;
2d7202bf
TG		 2612
2613 rtm = nlmsg_data(nlh);
1da177e4
LT		 2614 rtm->rtm_family = AF_INET6;
2615 rtm->rtm_dst_len = rt->rt6i_dst.plen;
2616 rtm->rtm_src_len = rt->rt6i_src.plen;
2617 rtm->rtm_tos = 0;
c71099ac 2618 if (rt->rt6i_table)
9e762a4a 2619 table = rt->rt6i_table->tb6_id;
c71099ac 2620 else
9e762a4a
PM		 2621 table = RT6_TABLE_UNSPEC;
2622 rtm->rtm_table = table;
c78679e8
DM		 2623 if (nla_put_u32(skb, RTA_TABLE, table))
2624 goto nla_put_failure;
ef2c7d7b
ND		 2625 if (rt->rt6i_flags & RTF_REJECT) {
2626 switch (rt->dst.error) {
2627 case -EINVAL:
2628 rtm->rtm_type = RTN_BLACKHOLE;
2629 break;
2630 case -EACCES:
2631 rtm->rtm_type = RTN_PROHIBIT;
2632 break;
b4949ab2
ND		 2633 case -EAGAIN:
2634 rtm->rtm_type = RTN_THROW;
2635 break;
ef2c7d7b
ND		 2636 default:
2637 rtm->rtm_type = RTN_UNREACHABLE;
2638 break;
2639 }
2640 }
38308473 2641 else if (rt->rt6i_flags & RTF_LOCAL)
ab79ad14 2642 rtm->rtm_type = RTN_LOCAL;
d1918542 2643 else if (rt->dst.dev && (rt->dst.dev->flags & IFF_LOOPBACK))
1da177e4
LT		 2644 rtm->rtm_type = RTN_LOCAL;
2645 else
2646 rtm->rtm_type = RTN_UNICAST;
2647 rtm->rtm_flags = 0;
2648 rtm->rtm_scope = RT_SCOPE_UNIVERSE;
2649 rtm->rtm_protocol = rt->rt6i_protocol;
38308473 2650 if (rt->rt6i_flags & RTF_DYNAMIC)
1da177e4 2651 rtm->rtm_protocol = RTPROT_REDIRECT;
f0396f60
DO		 2652 else if (rt->rt6i_flags & RTF_ADDRCONF) {
2653 if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ROUTEINFO))
2654 rtm->rtm_protocol = RTPROT_RA;
2655 else
2656 rtm->rtm_protocol = RTPROT_KERNEL;
2657 }
1da177e4 2658
38308473 2659 if (rt->rt6i_flags & RTF_CACHE)
1da177e4
LT		 2660 rtm->rtm_flags |= RTM_F_CLONED;
2661
2662 if (dst) {
c78679e8
DM		 2663 if (nla_put(skb, RTA_DST, 16, dst))
2664 goto nla_put_failure;
1ab1457c 2665 rtm->rtm_dst_len = 128;
1da177e4 2666 } else if (rtm->rtm_dst_len)
c78679e8
DM		 2667 if (nla_put(skb, RTA_DST, 16, &rt->rt6i_dst.addr))
2668 goto nla_put_failure;
1da177e4
LT		 2669#ifdef CONFIG_IPV6_SUBTREES
2670 if (src) {
c78679e8
DM		 2671 if (nla_put(skb, RTA_SRC, 16, src))
2672 goto nla_put_failure;
1ab1457c 2673 rtm->rtm_src_len = 128;
c78679e8
DM		 2674 } else if (rtm->rtm_src_len &&
2675 nla_put(skb, RTA_SRC, 16, &rt->rt6i_src.addr))
2676 goto nla_put_failure;
1da177e4 2677#endif
7bc570c8
YH		 2678 if (iif) {
2679#ifdef CONFIG_IPV6_MROUTE
2680 if (ipv6_addr_is_multicast(&rt->rt6i_dst.addr)) {
8229efda 2681 int err = ip6mr_get_route(net, skb, rtm, nowait);
7bc570c8
YH		 2682 if (err <= 0) {
2683 if (!nowait) {
2684 if (err == 0)
2685 return 0;
2686 goto nla_put_failure;
2687 } else {
2688 if (err == -EMSGSIZE)
2689 goto nla_put_failure;
2690 }
2691 }
2692 } else
2693#endif
c78679e8
DM		 2694 if (nla_put_u32(skb, RTA_IIF, iif))
2695 goto nla_put_failure;
7bc570c8 2696 } else if (dst) {
1da177e4 2697 struct in6_addr saddr_buf;
c78679e8
DM		 2698 if (ip6_route_get_saddr(net, rt, dst, 0, &saddr_buf) == 0 &&
2699 nla_put(skb, RTA_PREFSRC, 16, &saddr_buf))
2700 goto nla_put_failure;
1da177e4 2701 }
2d7202bf 2702
c3968a85
DW		 2703 if (rt->rt6i_prefsrc.plen) {
2704 struct in6_addr saddr_buf;
4e3fd7a0 2705 saddr_buf = rt->rt6i_prefsrc.addr;
c78679e8
DM		 2706 if (nla_put(skb, RTA_PREFSRC, 16, &saddr_buf))
2707 goto nla_put_failure;
c3968a85
DW		 2708 }
2709
defb3519 2710 if (rtnetlink_put_metrics(skb, dst_metrics_ptr(&rt->dst)) < 0)
2d7202bf
TG		 2711 goto nla_put_failure;
2712
dd0cbf29
YH		 2713 if (rt->rt6i_flags & RTF_GATEWAY) {
2714 if (nla_put(skb, RTA_GATEWAY, 16, &rt->rt6i_gateway) < 0)
94f826b8 2715 goto nla_put_failure;
94f826b8 2716 }
2d7202bf 2717
c78679e8
DM		 2718 if (rt->dst.dev &&
2719 nla_put_u32(skb, RTA_OIF, rt->dst.dev->ifindex))
2720 goto nla_put_failure;
2721 if (nla_put_u32(skb, RTA_PRIORITY, rt->rt6i_metric))
2722 goto nla_put_failure;
8253947e
LW		 2723
2724 expires = (rt->rt6i_flags & RTF_EXPIRES) ? rt->dst.expires - jiffies : 0;
69cdf8f9 2725
87a50699 2726 if (rtnl_put_cacheinfo(skb, &rt->dst, 0, expires, rt->dst.error) < 0)
e3703b3d 2727 goto nla_put_failure;
2d7202bf 2728
053c095a
JB		 2729 nlmsg_end(skb, nlh);
2730 return 0;
2d7202bf
TG		 2731
2732nla_put_failure:
26932566
PM		 2733 nlmsg_cancel(skb, nlh);
2734 return -EMSGSIZE;
1da177e4
LT		 2735}
2736
1b43af54 2737int rt6_dump_route(struct rt6_info *rt, void *p_arg)
1da177e4
LT		 2738{
2739 struct rt6_rtnl_dump_arg *arg = (struct rt6_rtnl_dump_arg *) p_arg;
2740 int prefix;
2741
2d7202bf
TG		 2742 if (nlmsg_len(arg->cb->nlh) >= sizeof(struct rtmsg)) {
2743 struct rtmsg *rtm = nlmsg_data(arg->cb->nlh);
1da177e4
LT		 2744 prefix = (rtm->rtm_flags & RTM_F_PREFIX) != 0;
2745 } else
2746 prefix = 0;
2747
191cd582
BH		 2748 return rt6_fill_node(arg->net,
2749 arg->skb, rt, NULL, NULL, 0, RTM_NEWROUTE,
15e47304 2750 NETLINK_CB(arg->cb->skb).portid, arg->cb->nlh->nlmsg_seq,
7bc570c8 2751 prefix, 0, NLM_F_MULTI);
1da177e4
LT		 2752}
2753
67ba4152 2754static int inet6_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr *nlh)
1da177e4 2755{
3b1e0a65 2756 struct net *net = sock_net(in_skb->sk);
ab364a6f
TG		 2757 struct nlattr *tb[RTA_MAX+1];
2758 struct rt6_info *rt;
1da177e4 2759 struct sk_buff *skb;
ab364a6f 2760 struct rtmsg *rtm;
4c9483b2 2761 struct flowi6 fl6;
72331bc0 2762 int err, iif = 0, oif = 0;
1da177e4 2763
ab364a6f
TG		 2764 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy);
2765 if (err < 0)
2766 goto errout;
1da177e4 2767
ab364a6f 2768 err = -EINVAL;
4c9483b2 2769 memset(&fl6, 0, sizeof(fl6));
1da177e4 2770
ab364a6f
TG		 2771 if (tb[RTA_SRC]) {
2772 if (nla_len(tb[RTA_SRC]) < sizeof(struct in6_addr))
2773 goto errout;
2774
4e3fd7a0 2775 fl6.saddr = *(struct in6_addr *)nla_data(tb[RTA_SRC]);
ab364a6f
TG		 2776 }
2777
2778 if (tb[RTA_DST]) {
2779 if (nla_len(tb[RTA_DST]) < sizeof(struct in6_addr))
2780 goto errout;
2781
4e3fd7a0 2782 fl6.daddr = *(struct in6_addr *)nla_data(tb[RTA_DST]);
ab364a6f
TG		 2783 }
2784
2785 if (tb[RTA_IIF])
2786 iif = nla_get_u32(tb[RTA_IIF]);
2787
2788 if (tb[RTA_OIF])
72331bc0 2789 oif = nla_get_u32(tb[RTA_OIF]);
1da177e4 2790
2e47b291
LC		 2791 if (tb[RTA_MARK])
2792 fl6.flowi6_mark = nla_get_u32(tb[RTA_MARK]);
2793
1da177e4
LT		 2794 if (iif) {
2795 struct net_device *dev;
72331bc0
SL		 2796 int flags = 0;
2797
5578689a 2798 dev = __dev_get_by_index(net, iif);
1da177e4
LT		 2799 if (!dev) {
2800 err = -ENODEV;
ab364a6f 2801 goto errout;
1da177e4 2802 }
72331bc0
SL		 2803
2804 fl6.flowi6_iif = iif;
2805
2806 if (!ipv6_addr_any(&fl6.saddr))
2807 flags |= RT6_LOOKUP_F_HAS_SADDR;
2808
2809 rt = (struct rt6_info *)ip6_route_input_lookup(net, dev, &fl6,
2810 flags);
2811 } else {
2812 fl6.flowi6_oif = oif;
2813
2814 rt = (struct rt6_info *)ip6_route_output(net, NULL, &fl6);
1da177e4
LT		 2815 }
2816
ab364a6f 2817 skb = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL);
38308473 2818 if (!skb) {
94e187c0 2819 ip6_rt_put(rt);
ab364a6f
TG		 2820 err = -ENOBUFS;
2821 goto errout;
2822 }
1da177e4 2823
ab364a6f
TG		 2824 /* Reserve room for dummy headers, this skb can pass
2825 through good chunk of routing engine.
2826 */
459a98ed 2827 skb_reset_mac_header(skb);
ab364a6f 2828 skb_reserve(skb, MAX_HEADER + sizeof(struct ipv6hdr));
1da177e4 2829
d8d1f30b 2830 skb_dst_set(skb, &rt->dst);
1da177e4 2831
4c9483b2 2832 err = rt6_fill_node(net, skb, rt, &fl6.daddr, &fl6.saddr, iif,
15e47304 2833 RTM_NEWROUTE, NETLINK_CB(in_skb).portid,
7bc570c8 2834 nlh->nlmsg_seq, 0, 0, 0);
1da177e4 2835 if (err < 0) {
ab364a6f
TG		 2836 kfree_skb(skb);
2837 goto errout;
1da177e4
LT		 2838 }
2839
15e47304 2840 err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).portid);
ab364a6f 2841errout:
1da177e4 2842 return err;
1da177e4
LT		 2843}
2844
86872cb5 2845void inet6_rt_notify(int event, struct rt6_info *rt, struct nl_info *info)
1da177e4
LT		 2846{
2847 struct sk_buff *skb;
5578689a 2848 struct net *net = info->nl_net;
528c4ceb
DL		 2849 u32 seq;
2850 int err;
2851
2852 err = -ENOBUFS;
38308473 2853 seq = info->nlh ? info->nlh->nlmsg_seq : 0;
86872cb5 2854
339bf98f 2855 skb = nlmsg_new(rt6_nlmsg_size(), gfp_any());
38308473 2856 if (!skb)
21713ebc
TG		 2857 goto errout;
2858
191cd582 2859 err = rt6_fill_node(net, skb, rt, NULL, NULL, 0,
15e47304 2860 event, info->portid, seq, 0, 0, 0);
26932566
PM		 2861 if (err < 0) {
2862 /* -EMSGSIZE implies BUG in rt6_nlmsg_size() */
2863 WARN_ON(err == -EMSGSIZE);
2864 kfree_skb(skb);
2865 goto errout;
2866 }
15e47304 2867 rtnl_notify(skb, net, info->portid, RTNLGRP_IPV6_ROUTE,
1ce85fe4
PNA		 2868 info->nlh, gfp_any());
2869 return;
21713ebc
TG		 2870errout:
2871 if (err < 0)
5578689a 2872 rtnl_set_sk_err(net, RTNLGRP_IPV6_ROUTE, err);
1da177e4
LT		 2873}
2874
8ed67789 2875static int ip6_route_dev_notify(struct notifier_block *this,
351638e7 2876 unsigned long event, void *ptr)
8ed67789 2877{
351638e7 2878 struct net_device *dev = netdev_notifier_info_to_dev(ptr);
c346dca1 2879 struct net *net = dev_net(dev);
8ed67789
DL		 2880
2881 if (event == NETDEV_REGISTER && (dev->flags & IFF_LOOPBACK)) {
d8d1f30b 2882 net->ipv6.ip6_null_entry->dst.dev = dev;
8ed67789
DL		 2883 net->ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(dev);
2884#ifdef CONFIG_IPV6_MULTIPLE_TABLES
d8d1f30b 2885 net->ipv6.ip6_prohibit_entry->dst.dev = dev;
8ed67789 2886 net->ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(dev);
d8d1f30b 2887 net->ipv6.ip6_blk_hole_entry->dst.dev = dev;
8ed67789
DL		 2888 net->ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(dev);
2889#endif
2890 }
2891
2892 return NOTIFY_OK;
2893}
2894
1da177e4
LT		 2895/*
2896 * /proc
2897 */
2898
2899#ifdef CONFIG_PROC_FS
2900
33120b30
AD		 2901static const struct file_operations ipv6_route_proc_fops = {
2902 .owner = THIS_MODULE,
2903 .open = ipv6_route_open,
2904 .read = seq_read,
2905 .llseek = seq_lseek,
8d2ca1d7 2906 .release = seq_release_net,
33120b30
AD		 2907};
2908
1da177e4
LT		 2909static int rt6_stats_seq_show(struct seq_file *seq, void *v)
2910{
69ddb805 2911 struct net *net = (struct net *)seq->private;
1da177e4 2912 seq_printf(seq, "%04x %04x %04x %04x %04x %04x %04x\n",
69ddb805
DL		 2913 net->ipv6.rt6_stats->fib_nodes,
2914 net->ipv6.rt6_stats->fib_route_nodes,
2915 net->ipv6.rt6_stats->fib_rt_alloc,
2916 net->ipv6.rt6_stats->fib_rt_entries,
2917 net->ipv6.rt6_stats->fib_rt_cache,
fc66f95c 2918 dst_entries_get_slow(&net->ipv6.ip6_dst_ops),
69ddb805 2919 net->ipv6.rt6_stats->fib_discarded_routes);
1da177e4
LT		 2920
2921 return 0;
2922}
2923
2924static int rt6_stats_seq_open(struct inode *inode, struct file *file)
2925{
de05c557 2926 return single_open_net(inode, file, rt6_stats_seq_show);
69ddb805
DL		 2927}
2928
9a32144e 2929static const struct file_operations rt6_stats_seq_fops = {
1da177e4
LT		 2930 .owner = THIS_MODULE,
2931 .open = rt6_stats_seq_open,
2932 .read = seq_read,
2933 .llseek = seq_lseek,
b6fcbdb4 2934 .release = single_release_net,
1da177e4
LT		 2935};
2936#endif /* CONFIG_PROC_FS */
2937
2938#ifdef CONFIG_SYSCTL
2939
1da177e4 2940static
fe2c6338 2941int ipv6_sysctl_rtcache_flush(struct ctl_table *ctl, int write,
1da177e4
LT		 2942 void __user *buffer, size_t *lenp, loff_t *ppos)
2943{
c486da34
LAG		 2944 struct net *net;
2945 int delay;
2946 if (!write)
1da177e4 2947 return -EINVAL;
c486da34
LAG		 2948
2949 net = (struct net *)ctl->extra1;
2950 delay = net->ipv6.sysctl.flush_delay;
2951 proc_dointvec(ctl, write, buffer, lenp, ppos);
2ac3ac8f 2952 fib6_run_gc(delay <= 0 ? 0 : (unsigned long)delay, net, delay > 0);
c486da34 2953 return 0;
1da177e4
LT		 2954}
2955
fe2c6338 2956struct ctl_table ipv6_route_table_template[] = {
1ab1457c 2957 {
1da177e4 2958 .procname = "flush",
4990509f 2959 .data = &init_net.ipv6.sysctl.flush_delay,
1da177e4 2960 .maxlen = sizeof(int),
89c8b3a1 2961 .mode = 0200,
6d9f239a 2962 .proc_handler = ipv6_sysctl_rtcache_flush
1da177e4
LT		 2963 },
2964 {
1da177e4 2965 .procname = "gc_thresh",
9a7ec3a9 2966 .data = &ip6_dst_ops_template.gc_thresh,
1da177e4
LT		 2967 .maxlen = sizeof(int),
2968 .mode = 0644,
6d9f239a 2969 .proc_handler = proc_dointvec,
1da177e4
LT		 2970 },
2971 {
1da177e4 2972 .procname = "max_size",
4990509f 2973 .data = &init_net.ipv6.sysctl.ip6_rt_max_size,
1da177e4
LT		 2974 .maxlen = sizeof(int),
2975 .mode = 0644,
6d9f239a 2976 .proc_handler = proc_dointvec,
1da177e4
LT		 2977 },
2978 {
1da177e4 2979 .procname = "gc_min_interval",
4990509f 2980 .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval,
1da177e4
LT		 2981 .maxlen = sizeof(int),
2982 .mode = 0644,
6d9f239a 2983 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 2984 },
2985 {
1da177e4 2986 .procname = "gc_timeout",
4990509f 2987 .data = &init_net.ipv6.sysctl.ip6_rt_gc_timeout,
1da177e4
LT		 2988 .maxlen = sizeof(int),
2989 .mode = 0644,
6d9f239a 2990 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 2991 },
2992 {
1da177e4 2993 .procname = "gc_interval",
4990509f 2994 .data = &init_net.ipv6.sysctl.ip6_rt_gc_interval,
1da177e4
LT		 2995 .maxlen = sizeof(int),
2996 .mode = 0644,
6d9f239a 2997 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 2998 },
2999 {
1da177e4 3000 .procname = "gc_elasticity",
4990509f 3001 .data = &init_net.ipv6.sysctl.ip6_rt_gc_elasticity,
1da177e4
LT		 3002 .maxlen = sizeof(int),
3003 .mode = 0644,
f3d3f616 3004 .proc_handler = proc_dointvec,
1da177e4
LT		 3005 },
3006 {
1da177e4 3007 .procname = "mtu_expires",
4990509f 3008 .data = &init_net.ipv6.sysctl.ip6_rt_mtu_expires,
1da177e4
LT		 3009 .maxlen = sizeof(int),
3010 .mode = 0644,
6d9f239a 3011 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 3012 },
3013 {
1da177e4 3014 .procname = "min_adv_mss",
4990509f 3015 .data = &init_net.ipv6.sysctl.ip6_rt_min_advmss,
1da177e4
LT		 3016 .maxlen = sizeof(int),
3017 .mode = 0644,
f3d3f616 3018 .proc_handler = proc_dointvec,
1da177e4
LT		 3019 },
3020 {
1da177e4 3021 .procname = "gc_min_interval_ms",
4990509f 3022 .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval,
1da177e4
LT		 3023 .maxlen = sizeof(int),
3024 .mode = 0644,
6d9f239a 3025 .proc_handler = proc_dointvec_ms_jiffies,
1da177e4 3026 },
f8572d8f 3027 { }
1da177e4
LT		 3028};
3029
2c8c1e72 3030struct ctl_table * __net_init ipv6_route_sysctl_init(struct net *net)
760f2d01
DL		 3031{
3032 struct ctl_table *table;
3033
3034 table = kmemdup(ipv6_route_table_template,
3035 sizeof(ipv6_route_table_template),
3036 GFP_KERNEL);
5ee09105
YH		 3037
3038 if (table) {
3039 table[0].data = &net->ipv6.sysctl.flush_delay;
c486da34 3040 table[0].extra1 = net;
86393e52 3041 table[1].data = &net->ipv6.ip6_dst_ops.gc_thresh;
5ee09105
YH		 3042 table[2].data = &net->ipv6.sysctl.ip6_rt_max_size;
3043 table[3].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval;
3044 table[4].data = &net->ipv6.sysctl.ip6_rt_gc_timeout;
3045 table[5].data = &net->ipv6.sysctl.ip6_rt_gc_interval;
3046 table[6].data = &net->ipv6.sysctl.ip6_rt_gc_elasticity;
3047 table[7].data = &net->ipv6.sysctl.ip6_rt_mtu_expires;
3048 table[8].data = &net->ipv6.sysctl.ip6_rt_min_advmss;
9c69fabe 3049 table[9].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval;
464dc801
EB		 3050
3051 /* Don't export sysctls to unprivileged users */
3052 if (net->user_ns != &init_user_ns)
3053 table[0].procname = NULL;
5ee09105
YH		 3054 }
3055
760f2d01
DL		 3056 return table;
3057}
1da177e4
LT		 3058#endif
3059
2c8c1e72 3060static int __net_init ip6_route_net_init(struct net *net)
cdb18761 3061{
633d424b 3062 int ret = -ENOMEM;
8ed67789 3063
86393e52
AD		 3064 memcpy(&net->ipv6.ip6_dst_ops, &ip6_dst_ops_template,
3065 sizeof(net->ipv6.ip6_dst_ops));
f2fc6a54 3066
fc66f95c
ED		 3067 if (dst_entries_init(&net->ipv6.ip6_dst_ops) < 0)
3068 goto out_ip6_dst_ops;
3069
8ed67789
DL		 3070 net->ipv6.ip6_null_entry = kmemdup(&ip6_null_entry_template,
3071 sizeof(*net->ipv6.ip6_null_entry),
3072 GFP_KERNEL);
3073 if (!net->ipv6.ip6_null_entry)
fc66f95c 3074 goto out_ip6_dst_entries;
d8d1f30b 3075 net->ipv6.ip6_null_entry->dst.path =
8ed67789 3076 (struct dst_entry *)net->ipv6.ip6_null_entry;
d8d1f30b 3077 net->ipv6.ip6_null_entry->dst.ops = &net->ipv6.ip6_dst_ops;
62fa8a84
DM		 3078 dst_init_metrics(&net->ipv6.ip6_null_entry->dst,
3079 ip6_template_metrics, true);
8ed67789
DL		 3080
3081#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3082 net->ipv6.ip6_prohibit_entry = kmemdup(&ip6_prohibit_entry_template,
3083 sizeof(*net->ipv6.ip6_prohibit_entry),
3084 GFP_KERNEL);
68fffc67
PZ		 3085 if (!net->ipv6.ip6_prohibit_entry)
3086 goto out_ip6_null_entry;
d8d1f30b 3087 net->ipv6.ip6_prohibit_entry->dst.path =
8ed67789 3088 (struct dst_entry *)net->ipv6.ip6_prohibit_entry;
d8d1f30b 3089 net->ipv6.ip6_prohibit_entry->dst.ops = &net->ipv6.ip6_dst_ops;
62fa8a84
DM		 3090 dst_init_metrics(&net->ipv6.ip6_prohibit_entry->dst,
3091 ip6_template_metrics, true);
8ed67789
DL		 3092
3093 net->ipv6.ip6_blk_hole_entry = kmemdup(&ip6_blk_hole_entry_template,
3094 sizeof(*net->ipv6.ip6_blk_hole_entry),
3095 GFP_KERNEL);
68fffc67
PZ		 3096 if (!net->ipv6.ip6_blk_hole_entry)
3097 goto out_ip6_prohibit_entry;
d8d1f30b 3098 net->ipv6.ip6_blk_hole_entry->dst.path =
8ed67789 3099 (struct dst_entry *)net->ipv6.ip6_blk_hole_entry;
d8d1f30b 3100 net->ipv6.ip6_blk_hole_entry->dst.ops = &net->ipv6.ip6_dst_ops;
62fa8a84
DM		 3101 dst_init_metrics(&net->ipv6.ip6_blk_hole_entry->dst,
3102 ip6_template_metrics, true);
8ed67789
DL		 3103#endif
3104
b339a47c
PZ		 3105 net->ipv6.sysctl.flush_delay = 0;
3106 net->ipv6.sysctl.ip6_rt_max_size = 4096;
3107 net->ipv6.sysctl.ip6_rt_gc_min_interval = HZ / 2;
3108 net->ipv6.sysctl.ip6_rt_gc_timeout = 60*HZ;
3109 net->ipv6.sysctl.ip6_rt_gc_interval = 30*HZ;
3110 net->ipv6.sysctl.ip6_rt_gc_elasticity = 9;
3111 net->ipv6.sysctl.ip6_rt_mtu_expires = 10*60*HZ;
3112 net->ipv6.sysctl.ip6_rt_min_advmss = IPV6_MIN_MTU - 20 - 40;
3113
6891a346
BT		 3114 net->ipv6.ip6_rt_gc_expire = 30*HZ;
3115
8ed67789
DL		 3116 ret = 0;
3117out:
3118 return ret;
f2fc6a54 3119
68fffc67
PZ		 3120#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3121out_ip6_prohibit_entry:
3122 kfree(net->ipv6.ip6_prohibit_entry);
3123out_ip6_null_entry:
3124 kfree(net->ipv6.ip6_null_entry);
3125#endif
fc66f95c
ED		 3126out_ip6_dst_entries:
3127 dst_entries_destroy(&net->ipv6.ip6_dst_ops);
f2fc6a54 3128out_ip6_dst_ops:
f2fc6a54 3129 goto out;
cdb18761
DL		 3130}
3131
2c8c1e72 3132static void __net_exit ip6_route_net_exit(struct net *net)
cdb18761 3133{
8ed67789
DL		 3134 kfree(net->ipv6.ip6_null_entry);
3135#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3136 kfree(net->ipv6.ip6_prohibit_entry);
3137 kfree(net->ipv6.ip6_blk_hole_entry);
3138#endif
41bb78b4 3139 dst_entries_destroy(&net->ipv6.ip6_dst_ops);
cdb18761
DL		 3140}
3141
d189634e
TG		 3142static int __net_init ip6_route_net_init_late(struct net *net)
3143{
3144#ifdef CONFIG_PROC_FS
d4beaa66
G		 3145 proc_create("ipv6_route", 0, net->proc_net, &ipv6_route_proc_fops);
3146 proc_create("rt6_stats", S_IRUGO, net->proc_net, &rt6_stats_seq_fops);
d189634e
TG		 3147#endif
3148 return 0;
3149}
3150
3151static void __net_exit ip6_route_net_exit_late(struct net *net)
3152{
3153#ifdef CONFIG_PROC_FS
ece31ffd
G		 3154 remove_proc_entry("ipv6_route", net->proc_net);
3155 remove_proc_entry("rt6_stats", net->proc_net);
d189634e
TG		 3156#endif
3157}
3158
cdb18761
DL		 3159static struct pernet_operations ip6_route_net_ops = {
3160 .init = ip6_route_net_init,
3161 .exit = ip6_route_net_exit,
3162};
3163
c3426b47
DM		 3164static int __net_init ipv6_inetpeer_init(struct net *net)
3165{
3166 struct inet_peer_base *bp = kmalloc(sizeof(*bp), GFP_KERNEL);
3167
3168 if (!bp)
3169 return -ENOMEM;
3170 inet_peer_base_init(bp);
3171 net->ipv6.peers = bp;
3172 return 0;
3173}
3174
3175static void __net_exit ipv6_inetpeer_exit(struct net *net)
3176{
3177 struct inet_peer_base *bp = net->ipv6.peers;
3178
3179 net->ipv6.peers = NULL;
56a6b248 3180 inetpeer_invalidate_tree(bp);
c3426b47
DM		 3181 kfree(bp);
3182}
3183
2b823f72 3184static struct pernet_operations ipv6_inetpeer_ops = {
c3426b47
DM		 3185 .init = ipv6_inetpeer_init,
3186 .exit = ipv6_inetpeer_exit,
3187};
3188
d189634e
TG		 3189static struct pernet_operations ip6_route_net_late_ops = {
3190 .init = ip6_route_net_init_late,
3191 .exit = ip6_route_net_exit_late,
3192};
3193
8ed67789
DL		 3194static struct notifier_block ip6_route_dev_notifier = {
3195 .notifier_call = ip6_route_dev_notify,
3196 .priority = 0,
3197};
3198
433d49c3 3199int __init ip6_route_init(void)
1da177e4 3200{
433d49c3
DL		 3201 int ret;
3202
9a7ec3a9
DL		 3203 ret = -ENOMEM;
3204 ip6_dst_ops_template.kmem_cachep =
e5d679f3 3205 kmem_cache_create("ip6_dst_cache", sizeof(struct rt6_info), 0,
f845ab6b 3206 SLAB_HWCACHE_ALIGN, NULL);
9a7ec3a9 3207 if (!ip6_dst_ops_template.kmem_cachep)
c19a28e1 3208 goto out;
14e50e57 3209
fc66f95c 3210 ret = dst_entries_init(&ip6_dst_blackhole_ops);
8ed67789 3211 if (ret)
bdb3289f 3212 goto out_kmem_cache;
bdb3289f 3213
c3426b47
DM		 3214 ret = register_pernet_subsys(&ipv6_inetpeer_ops);
3215 if (ret)
e8803b6c 3216 goto out_dst_entries;
2a0c451a 3217
7e52b33b
DM		 3218 ret = register_pernet_subsys(&ip6_route_net_ops);
3219 if (ret)
3220 goto out_register_inetpeer;
c3426b47 3221
5dc121e9
AE		 3222 ip6_dst_blackhole_ops.kmem_cachep = ip6_dst_ops_template.kmem_cachep;
3223
8ed67789
DL		 3224 /* Registering of the loopback is done before this portion of code,
3225 * the loopback reference in rt6_info will not be taken, do it
3226 * manually for init_net */
d8d1f30b 3227 init_net.ipv6.ip6_null_entry->dst.dev = init_net.loopback_dev;
8ed67789
DL		 3228 init_net.ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
3229 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
d8d1f30b 3230 init_net.ipv6.ip6_prohibit_entry->dst.dev = init_net.loopback_dev;
8ed67789 3231 init_net.ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
d8d1f30b 3232 init_net.ipv6.ip6_blk_hole_entry->dst.dev = init_net.loopback_dev;
8ed67789
DL		 3233 init_net.ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
3234 #endif
e8803b6c 3235 ret = fib6_init();
433d49c3 3236 if (ret)
8ed67789 3237 goto out_register_subsys;
433d49c3 3238
433d49c3
DL		 3239 ret = xfrm6_init();
3240 if (ret)
e8803b6c 3241 goto out_fib6_init;
c35b7e72 3242
433d49c3
DL		 3243 ret = fib6_rules_init();
3244 if (ret)
3245 goto xfrm6_init;
7e5449c2 3246
d189634e
TG		 3247 ret = register_pernet_subsys(&ip6_route_net_late_ops);
3248 if (ret)
3249 goto fib6_rules_init;
3250
433d49c3 3251 ret = -ENOBUFS;
c7ac8679
GR		 3252 if (__rtnl_register(PF_INET6, RTM_NEWROUTE, inet6_rtm_newroute, NULL, NULL) ||
3253 __rtnl_register(PF_INET6, RTM_DELROUTE, inet6_rtm_delroute, NULL, NULL) ||
3254 __rtnl_register(PF_INET6, RTM_GETROUTE, inet6_rtm_getroute, NULL, NULL))
d189634e 3255 goto out_register_late_subsys;
c127ea2c 3256
8ed67789 3257 ret = register_netdevice_notifier(&ip6_route_dev_notifier);
cdb18761 3258 if (ret)
d189634e 3259 goto out_register_late_subsys;
8ed67789 3260
433d49c3
DL		 3261out:
3262 return ret;
3263
d189634e
TG		 3264out_register_late_subsys:
3265 unregister_pernet_subsys(&ip6_route_net_late_ops);
433d49c3 3266fib6_rules_init:
433d49c3
DL		 3267 fib6_rules_cleanup();
3268xfrm6_init:
433d49c3 3269 xfrm6_fini();
2a0c451a
TG		 3270out_fib6_init:
3271 fib6_gc_cleanup();
8ed67789
DL		 3272out_register_subsys:
3273 unregister_pernet_subsys(&ip6_route_net_ops);
7e52b33b
DM		 3274out_register_inetpeer:
3275 unregister_pernet_subsys(&ipv6_inetpeer_ops);
fc66f95c
ED		 3276out_dst_entries:
3277 dst_entries_destroy(&ip6_dst_blackhole_ops);
433d49c3 3278out_kmem_cache:
f2fc6a54 3279 kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep);
433d49c3 3280 goto out;
1da177e4
LT		 3281}
3282
3283void ip6_route_cleanup(void)
3284{
8ed67789 3285 unregister_netdevice_notifier(&ip6_route_dev_notifier);
d189634e 3286 unregister_pernet_subsys(&ip6_route_net_late_ops);
101367c2 3287 fib6_rules_cleanup();
1da177e4 3288 xfrm6_fini();
1da177e4 3289 fib6_gc_cleanup();
c3426b47 3290 unregister_pernet_subsys(&ipv6_inetpeer_ops);
8ed67789 3291 unregister_pernet_subsys(&ip6_route_net_ops);
41bb78b4 3292 dst_entries_destroy(&ip6_dst_blackhole_ops);
f2fc6a54 3293 kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep);
1da177e4 3294}
Mirror https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git